1
00:00:00,000 --> 00:00:07,000
Okay, yeah, the secret to life is to hit record slash send tweet.

2
00:00:07,000 --> 00:00:09,000
You can always have it.

3
00:00:09,000 --> 00:00:11,000
No, I don't edit, ever.

4
00:00:11,000 --> 00:00:12,000
Oh, fine.

5
00:00:12,000 --> 00:00:14,000
No, it's too much work.

6
00:00:14,000 --> 00:00:18,000
Just hit record and pray and see what happens.

7
00:00:18,000 --> 00:00:19,000
Prompt and pray.

8
00:00:19,000 --> 00:00:20,000
Exactly.

9
00:00:20,000 --> 00:00:21,000
Prompt and pray.

10
00:00:21,000 --> 00:00:22,000
That's the...

11
00:00:22,000 --> 00:00:24,000
With emphasis on pray.

12
00:00:24,000 --> 00:00:25,000
Because it's just...

13
00:00:25,000 --> 00:00:28,000
It never gets it right the first time.

14
00:00:28,000 --> 00:00:30,000
But Pablo is going to fix that.

15
00:00:30,000 --> 00:00:37,000
I mean, that was kind of the goal of the last couple of weeks, is to teach the agents how

16
00:00:37,000 --> 00:00:39,000
to build an ostrich properly.

17
00:00:39,000 --> 00:00:40,000
Yeah.

18
00:00:40,000 --> 00:00:41,000
I think we're really close.

19
00:00:41,000 --> 00:00:43,000
What's your guess?

20
00:00:43,000 --> 00:00:46,000
One shouting like a DVM or something.

21
00:00:46,000 --> 00:00:47,000
I think that...

22
00:00:47,000 --> 00:00:50,000
I mean, yeah, we are definitely into something.

23
00:00:50,000 --> 00:00:57,000
But I believe that, you know, these publicity cooks, you know, how they work.

24
00:00:57,000 --> 00:01:02,000
When some new technology appears and the publicity cook that they have where...

25
00:01:02,000 --> 00:01:03,000
Yeah, yeah.

26
00:01:03,000 --> 00:01:04,000
Now we're in the hype phase.

27
00:01:04,000 --> 00:01:05,000
Yeah.

28
00:01:05,000 --> 00:01:06,000
We're in the hype phase.

29
00:01:06,000 --> 00:01:08,000
We haven't reached the top of that.

30
00:01:08,000 --> 00:01:14,000
And then when we reach the top, it's going to be a moment of reflection.

31
00:01:14,000 --> 00:01:16,000
The trough of disillusionment, right?

32
00:01:16,000 --> 00:01:18,000
That's what it's called, I think.

33
00:01:18,000 --> 00:01:20,000
Everyone's like, oh, everything sucks.

34
00:01:20,000 --> 00:01:21,000
Yeah.

35
00:01:21,000 --> 00:01:22,000
Bitcoin is the ostrich.

36
00:01:22,000 --> 00:01:23,000
A lot of them are stupid.

37
00:01:23,000 --> 00:01:24,000
Exactly.

38
00:01:25,000 --> 00:01:26,000
Nothing works.

39
00:01:26,000 --> 00:01:29,000
I think it's something that you can...

40
00:01:29,000 --> 00:01:36,000
I mean, I like to be informed and to read articles, stay up-to-date news and so on.

41
00:01:36,000 --> 00:01:39,000
Did you read the 2027 one?

42
00:01:39,000 --> 00:01:40,000
Yeah, 2027.

43
00:01:40,000 --> 00:01:41,000
No.

44
00:01:41,000 --> 00:01:43,000
Hyper bullish super intelligence like tomorrow.

45
00:01:43,000 --> 00:01:44,000
No.

46
00:01:44,000 --> 00:01:45,000
No, I haven't read it completely.

47
00:01:45,000 --> 00:01:49,000
I read like an overview of it.

48
00:01:49,000 --> 00:01:52,000
But I haven't read it completely.

49
00:01:52,000 --> 00:01:59,000
I've been reading a lot of articles lately about security implications of these new protocols,

50
00:01:59,000 --> 00:02:02,000
like I-NCP or the AI.

51
00:02:02,000 --> 00:02:03,000
Yeah.

52
00:02:03,000 --> 00:02:04,000
They will be agents.

53
00:02:04,000 --> 00:02:05,000
Yeah.

54
00:02:05,000 --> 00:02:06,000
Yeah.

55
00:02:06,000 --> 00:02:07,000
Yeah.

56
00:02:07,000 --> 00:02:08,000
Yeah.

57
00:02:08,000 --> 00:02:09,000
Yeah.

58
00:02:09,000 --> 00:02:10,000
Yeah.

59
00:02:10,000 --> 00:02:11,000
Yeah.

60
00:02:11,000 --> 00:02:12,000
Yeah.

61
00:02:12,000 --> 00:02:13,000
Yeah.

62
00:02:13,000 --> 00:02:14,000
Yeah.

63
00:02:14,000 --> 00:02:15,000
Yeah.

64
00:02:15,000 --> 00:02:16,000
All sorts of possible attacks that we are not advising right now because, you know, it's so new.

65
00:02:16,000 --> 00:02:17,000
Yeah.

66
00:02:17,000 --> 00:02:18,000
No one knows.

67
00:02:18,000 --> 00:02:19,000
Well, it's not that violent.

68
00:02:20,000 --> 00:02:21,000
Yeah.

69
00:02:21,000 --> 00:02:22,000
Yeah.

70
00:02:22,000 --> 00:02:23,000
It's a nice promise, actually.

71
00:02:23,000 --> 00:02:28,000
So I think that all of these programatics that we are advising right now are something

72
00:02:28,000 --> 00:02:36,000
that people or researchers and security experts are putting on the public with all these articles.

73
00:02:36,000 --> 00:02:39,000
We are not suffering all of that, yet.

74
00:02:39,000 --> 00:02:45,000
Because these tools are so new and these protocols are so new and it really...

75
00:02:45,000 --> 00:02:51,880
attach it to a very niche part of the population, it's also not everyone's like playing around

76
00:02:51,880 --> 00:02:56,840
with it yet, it's not widely embedded everywhere yet, but it will be very soon I feel like,

77
00:02:56,840 --> 00:03:01,160
I mean mostly it's mostly the good guys so to speak playing around with it now you know like

78
00:03:02,040 --> 00:03:07,960
people most exposed to all these tools are basically AI researchers and co-workers and so on and

79
00:03:07,960 --> 00:03:13,480
so forth, but I think this will change very very quickly. Yeah definitely I think that we are

80
00:03:13,480 --> 00:03:22,280
going to reach that critical mass, but like before that it's important that we as like a kind of

81
00:03:24,360 --> 00:03:34,040
explorers of this new world, we can abide the challenges of this new tools. I mean my advice to

82
00:03:34,040 --> 00:03:42,520
everyone would be getting new laptop, a dedicated vibe coding machine, like get dedicated hardware,

83
00:03:42,600 --> 00:03:48,120
don't put anything of importance on it and just like switch it to your low mode, that would be

84
00:03:48,120 --> 00:03:56,280
my best. That would be my best advice, use QBOs and this virtualized everything. Is that what

85
00:03:56,280 --> 00:04:01,880
you do currently? No, okay, you have one machine, let's go down here otherwise we'll hit traffic,

86
00:04:02,680 --> 00:04:11,400
you have one machine and you just do a lot of things. I know, I'm not Jolo, I'm very cautious,

87
00:04:11,480 --> 00:04:18,600
you also use a different setup the most. You use Goose a lot as well, right? Yeah,

88
00:04:18,600 --> 00:04:24,520
I'm just experimenting with the power tools, you know, as I'm developing this DBnCP stuff,

89
00:04:24,520 --> 00:04:31,480
I'm playing a lot with NCP and so on. I normally don't use this as a consumer, I use it as a

90
00:04:31,480 --> 00:04:39,080
developer, so I try to find the best tools for that and Goose is amazing indeed, because for

91
00:04:39,080 --> 00:04:45,560
example, necessarily when I was introducing this dynamic discovering of tools, MCP supports

92
00:04:45,560 --> 00:04:51,880
these notifications from the server to the client where the server can notify to the client that

93
00:04:51,880 --> 00:04:58,440
the list of tools changed and get updated. For example, I was testing in use with Corsair or

94
00:04:58,440 --> 00:05:06,440
other clients and the only one that works was Goose. Okay, well, did you use Roocode also?

95
00:05:07,320 --> 00:05:14,440
No, I haven't tested yet. Okay. Were you part of the workshop with Paolo? Do you have a setup?

96
00:05:14,440 --> 00:05:19,720
Like in theory, do you have a working setup? Yeah, okay. Yeah, yeah, yeah, because that's like

97
00:05:19,720 --> 00:05:26,360
it's very configurable, I would be surprised if we couldn't get the updated tools and stuff.

98
00:05:26,360 --> 00:05:32,520
I really believe that the security measures that we are going to see in the incoming

99
00:05:32,600 --> 00:05:41,640
weeks, months, years are going to be in different levels, but one of them, one of the most important

100
00:05:41,640 --> 00:05:50,280
parts is going to be in the client level where you as a user execute tools from MCP servers and

101
00:05:50,280 --> 00:05:59,560
so on and the client should provide you enough, like a fishers to really understand what you are

102
00:05:59,560 --> 00:06:05,880
doing or edit the tools that you are executing. Yeah, and that's going to be a big security

103
00:06:08,360 --> 00:06:12,920
advantage. I think that's why when an Oscar is awesome in that regard and Hazard showed some

104
00:06:12,920 --> 00:06:17,000
of that off as well, because if you're doing something in an Oscar world, you basically,

105
00:06:18,840 --> 00:06:23,720
if you're using an external signer like Amber, it can't really do anything destructive in your

106
00:06:23,720 --> 00:06:29,000
name. Like you always have the executive power on your side just with the signer alone and it's

107
00:06:29,000 --> 00:06:35,320
like whatever your agent is doing and whatever tool call is made, Amber will always pop up and

108
00:06:35,320 --> 00:06:38,920
be like, okay, do you want to sign this event? Like if you have it configured that way, but I think

109
00:06:38,920 --> 00:06:46,920
that's super powerful. That's like a super interesting approach slash paradigm like the most other

110
00:06:46,920 --> 00:06:51,960
systems don't have, because usually it's all or nothing. It's like here's my API key and now you

111
00:06:51,960 --> 00:06:57,400
act on my behalf basically. And with the delegation possibilities of an Oscar, I think it's super

112
00:06:57,480 --> 00:07:02,360
interesting that you can, that you just can't have an agent and basically it can't really do

113
00:07:03,160 --> 00:07:07,960
anything destructive in that sense. So that's, I think that's an interesting side effect.

114
00:07:07,960 --> 00:07:12,680
Like it always hits you, you're a proverb, right? Like a disannoying window.

115
00:07:13,240 --> 00:07:20,040
Yeah, yeah, I mean, sure. I mean, that's why most people go just like, yeah, yeah, always,

116
00:07:20,040 --> 00:07:25,240
you know, always sign that. It's fine. Yeah, let's make a run for it. We can also, yeah.

117
00:07:26,200 --> 00:07:33,080
Oh, fuck. Oh, no, let's go. Breaking the law. Don't die traffic. Jesus rolled over by car.

118
00:07:34,440 --> 00:07:37,480
We made it. We made it. And we stay alive.

119
00:07:40,440 --> 00:07:45,400
You know, I need to keep you on your toes. Yeah. Everyone.

120
00:07:45,960 --> 00:07:49,800
Yeah, I think that, you know,

121
00:07:51,480 --> 00:07:56,120
Jolo mode is going to be something outdated so soon.

122
00:07:56,840 --> 00:08:02,280
Why? Because I don't believe that it's going to stay for longer because when more people start

123
00:08:02,280 --> 00:08:11,640
to use these new technologies and so on, we are not going to have that freedom to use execute

124
00:08:11,640 --> 00:08:16,040
everything in Jolo mode because they are going to be more exploit and more people want them to

125
00:08:16,040 --> 00:08:22,520
exploit. Oh, you think so? You think people will switch back from Jolo mode? I highly doubt it.

126
00:08:22,520 --> 00:08:26,840
No, I think that once you're in Jolo mode, once you're in Jolo mode, you never go back.

127
00:08:26,840 --> 00:08:31,960
I think that there is no going back. I think that they are different, different levels of Jolo mode.

128
00:08:31,960 --> 00:08:38,120
Like there is the reckless Jolo mode, but you go, yeah, yeah, sure. Totally unknown.

129
00:08:38,120 --> 00:08:43,000
I mean, Ru already has this where it's like, you know, you have more fine-grained permissions,

130
00:08:43,000 --> 00:08:46,600
but like it can always read a file, for example, but it can't always delete a file,

131
00:08:46,600 --> 00:08:52,600
those stuff like that. Yeah, yeah, yeah. But definitely the conception that we have about

132
00:08:52,600 --> 00:08:59,080
these tools have to change because it's impossible that we keep the same mentality that we have now

133
00:08:59,080 --> 00:09:04,520
for the future of these technologies because there are going to be so many attack vectors

134
00:09:05,000 --> 00:09:10,840
and people are going to be so compromised that I think that it's impossible to stay

135
00:09:10,840 --> 00:09:17,000
in the same level of Jolo mode. Yeah, I think I discussed this with Pablo already a little bit.

136
00:09:17,000 --> 00:09:22,920
It's what was like, everyone will run into this problem. Like, you know, what agent to trust,

137
00:09:22,920 --> 00:09:28,840
what MCP tools to trust, what kind of, I don't know, even code snippets to trust those kind of things.

138
00:09:28,840 --> 00:09:33,800
And I think Noster really has a unique position there because we do have identity

139
00:09:33,800 --> 00:09:40,360
and rep of trust baked in. So it's kind of, you know, trust everyone from my kind of closest

140
00:09:40,360 --> 00:09:45,480
circle on Noster. Then it's like, you know, I whitelist a couple of N-pups and I just trust

141
00:09:45,480 --> 00:09:50,200
whatever they sign off on or those kinds of things. Everyone else will kind of have to

142
00:09:50,200 --> 00:09:54,440
rebuild this from scratch in some sense. And they will, like, I mean, you know, there's many

143
00:09:55,000 --> 00:09:58,440
instances where this was done in the past, like, I don't know, like steam in the gaming world,

144
00:09:58,440 --> 00:10:02,920
rebuild the social graph just for like gaming, for example, you know. So they will definitely

145
00:10:03,800 --> 00:10:08,600
do that. And some of the companies will also succeed and so on. But I still think,

146
00:10:10,120 --> 00:10:13,560
I mean, we talked about this a lot the last couple of weeks, but everyone will run into

147
00:10:14,280 --> 00:10:18,840
those limits and those issues. And one of the limits is kind of web of trust. Like, one of the

148
00:10:18,840 --> 00:10:23,080
things you will run into is like, again, the trust question. The other one is the micropayment

149
00:10:23,080 --> 00:10:27,880
question. And I think, again, like, Noster, it's perfect for that. I would add the, the

150
00:10:27,960 --> 00:10:34,120
very family. Yeah. Oh, that that as well. Yeah, yeah. Yes. Yeah, which comes completely. I don't

151
00:10:34,120 --> 00:10:38,200
even think about it anymore because it so comes for free and then exercise. Like, everything is

152
00:10:38,200 --> 00:10:43,400
signed. So you don't really need to manually verify in that sense. Yeah. Yeah. That's true.

153
00:10:43,400 --> 00:10:51,480
I was, I was looking the this name specification for the eight to eight protocol. I was, I think I

154
00:10:51,480 --> 00:10:58,280
was talking to you about this the other day. But I was, I was watching this specification for

155
00:10:58,280 --> 00:11:04,440
the new Google protocol that eight to eight agent to agent. Yeah. Oh, yeah. Yeah. Yeah. They're trying

156
00:11:04,440 --> 00:11:09,960
to build a more complicated version of Noster basically. It's fucking awful. Yeah. It's a,

157
00:11:09,960 --> 00:11:17,800
the JSON specification is a 2000 lines JSON schema where there are a lot of nested objects and

158
00:11:18,760 --> 00:11:26,280
authentication method. Yeah. Yeah. It's fucking hell. Yeah. If you, if you use probably JW and

159
00:11:26,280 --> 00:11:31,960
JW SS all the way down in one big JSON object, I can guess how this looks like, but I don't want

160
00:11:31,960 --> 00:11:40,040
to look, I don't even want to look. It's not good to see. It's even worth to, to work with.

161
00:11:40,920 --> 00:11:45,400
With that. Yeah. I mean, it's safe to say that they will go down the path and maybe, you know,

162
00:11:45,400 --> 00:11:49,960
at some point in time in the future in a couple of years, they might turn around.

163
00:11:52,360 --> 00:11:56,360
If our thesis is correct and the Noster marketplace, so to speak,

164
00:11:56,680 --> 00:12:02,920
marketplace for agents and thinks is, is, works well enough than, than they might be

165
00:12:02,920 --> 00:12:07,400
grudgingly adopted later on. I mean, we made this comparison in the last couple of weeks, a lot with,

166
00:12:07,960 --> 00:12:13,960
you know, like how windows had to be grudgingly integrate Linux into the system, you know,

167
00:12:14,600 --> 00:12:20,280
all those kinds of things where it's just, that, that was thought impossible, like 20 years ago or

168
00:12:20,280 --> 00:12:26,520
so. And, and the open, it's going to be like the minimalistic open kind of, I don't know,

169
00:12:27,400 --> 00:12:33,160
you know, Unix ethos system just one out in the end, basically, and one out big time.

170
00:12:33,160 --> 00:12:37,960
So yeah, that's how I see this going. But what was your motivation, by the way, to

171
00:12:38,520 --> 00:12:41,080
start working on, on TV, MCP?

172
00:12:41,960 --> 00:12:42,920
Not them.

173
00:12:42,920 --> 00:12:43,320
Yeah.

174
00:12:43,320 --> 00:12:45,960
So we have to thank Tamadoches for this.

175
00:12:45,960 --> 00:12:46,440
Oh yeah.

176
00:12:46,440 --> 00:12:51,320
We have to thank, thank Tamadoches for so many things, but this is one of these things.

177
00:12:51,320 --> 00:12:52,280
We have to think.

178
00:12:52,280 --> 00:12:53,320
Awesome. I didn't know that.

179
00:12:54,280 --> 00:12:57,560
Yeah, yeah, yeah. No, in very few people will know what totem is, by the way.

180
00:12:57,560 --> 00:12:57,720
Yeah.

181
00:12:57,720 --> 00:12:59,960
Maybe explain it in a minute or two.

182
00:12:59,960 --> 00:13:03,480
Totem, totem is a, it's a project that's working with sandwich.

183
00:13:04,120 --> 00:13:09,000
We are creating some sort of experience on top of the data and nozzle.

184
00:13:09,720 --> 00:13:12,440
So you want to, the turn realized into Tamadoches, basically.

185
00:13:12,440 --> 00:13:13,080
Yeah, basically.

186
00:13:13,080 --> 00:13:13,640
Yeah.

187
00:13:13,640 --> 00:13:14,840
In the short years.

188
00:13:15,400 --> 00:13:20,120
But the other day was revisiting some notes that we have for, for totem.

189
00:13:21,560 --> 00:13:27,000
I saw the line that kind of give me the inspiration to work on the VNCP and was

190
00:13:27,560 --> 00:13:29,960
use DVM as a wrapper for NCP.

191
00:13:29,960 --> 00:13:30,520
Yeah.

192
00:13:30,520 --> 00:13:32,120
I think that was, that was the line.

193
00:13:32,120 --> 00:13:33,560
How long ago was it?

194
00:13:33,560 --> 00:13:34,360
A month ago.

195
00:13:34,360 --> 00:13:34,840
Yeah.

196
00:13:34,840 --> 00:13:36,920
Just before it started, it says, several in the end.

197
00:13:38,040 --> 00:13:39,000
That's funny.

198
00:13:39,000 --> 00:13:39,960
Life moves fast.

199
00:13:39,960 --> 00:13:40,440
Yeah.

200
00:13:40,440 --> 00:13:41,320
Life moves fast.

201
00:13:41,320 --> 00:13:42,040
Yeah.

202
00:13:42,040 --> 00:13:44,600
I mean, the whole NCP protocol is very young.

203
00:13:44,600 --> 00:13:45,160
Yeah, it is.

204
00:13:45,160 --> 00:13:45,320
It is.

205
00:13:45,320 --> 00:13:45,800
It is.

206
00:13:46,360 --> 00:13:53,720
To me, it's just a natural fit to use Noster to look up MCP tools and to look up.

207
00:13:53,720 --> 00:13:55,320
Like, look up all the things, basically.

208
00:13:55,320 --> 00:13:56,040
Look up.

209
00:13:56,040 --> 00:13:58,920
And that's what Pablo is building out right now.

210
00:13:58,920 --> 00:14:02,760
I'm going to share the tutorial again, but it's basically,

211
00:14:04,040 --> 00:14:08,600
what you want to do is you have these specialized agents and you want to kind of box them in.

212
00:14:08,600 --> 00:14:12,120
Because the main issue right now is that like, if the agents aren't constrained,

213
00:14:12,120 --> 00:14:14,680
they just go wild and stuff breaks all the time and so on.

214
00:14:14,680 --> 00:14:15,960
And they're also not Noster experts.

215
00:14:15,960 --> 00:14:18,840
They're like, not, agents aren't really experts in that sense.

216
00:14:18,840 --> 00:14:24,920
So, most large models in particular are generalists and so on.

217
00:14:25,320 --> 00:14:29,800
And so, the way to box them in is like, first of all, with the MCP tools,

218
00:14:29,800 --> 00:14:31,400
like what they can do, basically.

219
00:14:31,400 --> 00:14:36,520
But also, in terms of like, how do you use, I mean, I always use Pablo's example,

220
00:14:36,520 --> 00:14:40,520
how do you use NDK, you know, like, and you have various code snippets, like,

221
00:14:40,520 --> 00:14:43,080
okay, this is basically the documentation for you.

222
00:14:43,080 --> 00:14:45,000
This is how you do a profile lookup.

223
00:14:45,000 --> 00:14:47,960
This is how you do NIP-60, NIP-61, whatever.

224
00:14:47,960 --> 00:14:49,800
You know, this is how you sap someone.

225
00:14:49,800 --> 00:14:53,960
And then, then the LLMs aren't confused on how to do that.

226
00:14:53,960 --> 00:14:57,160
And the success rate of doing it in that way.

227
00:14:57,160 --> 00:14:58,680
And of course, you know, like, all this stuff,

228
00:14:58,680 --> 00:15:00,280
you can use Noster to look it up.

229
00:15:00,280 --> 00:15:02,120
Like, you look up the code snippets with Noster,

230
00:15:02,120 --> 00:15:05,240
you look up the MCP tools with Noster, all those kinds of things.

231
00:15:05,240 --> 00:15:09,480
And I basically, I used it myself a little bit as well,

232
00:15:09,480 --> 00:15:11,480
but I mostly watched over Pablo's shoulder.

233
00:15:11,480 --> 00:15:14,040
And the funny thing is, like, with the snippet approach, for example,

234
00:15:14,040 --> 00:15:17,080
if you fix an issue once, like, the LLM is always confused about,

235
00:15:17,080 --> 00:15:20,040
I don't know, like, even stupid things, like fetching some data.

236
00:15:20,040 --> 00:15:23,960
Because millions of answers on Stack Overflow

237
00:15:23,960 --> 00:15:26,360
were always like, OK, you query this REST API yet, so.

238
00:15:26,360 --> 00:15:28,520
And that's, like, in the Noster world, it's nonsensical.

239
00:15:28,520 --> 00:15:31,480
And it will always try to fetch something from an API and so on.

240
00:15:31,480 --> 00:15:34,760
And then you fix it once, and you say, OK, no, this is the way to fetch it.

241
00:15:34,760 --> 00:15:36,280
And it will never make that mistake again.

242
00:15:36,280 --> 00:15:37,560
And that's just so amazing to me.

243
00:15:37,560 --> 00:15:38,840
Like, that's so cool.

244
00:15:38,840 --> 00:15:43,880
And of course, with the NCP tools that we have now,

245
00:15:43,880 --> 00:15:45,800
it's like, basically, while you're coding,

246
00:15:45,800 --> 00:15:47,080
you can publish these code snippets.

247
00:15:47,080 --> 00:15:50,200
It's like, OK, you just made this mistake.

248
00:15:50,200 --> 00:15:52,440
Here is how you do it in a Noster way.

249
00:15:52,440 --> 00:15:54,280
Create a snippet, publish it to Noster,

250
00:15:54,280 --> 00:15:56,600
and don't confuse yourself anymore going forward.

251
00:15:56,600 --> 00:15:57,400
And it actually works.

252
00:15:57,400 --> 00:15:58,840
It's really amazing, so.

253
00:15:58,840 --> 00:16:02,520
Yeah, I think that the important part of this is that

254
00:16:02,520 --> 00:16:07,240
we have something like LLMs that are kind of black boxes,

255
00:16:07,240 --> 00:16:10,280
that are stateless, that doesn't have any concept of context.

256
00:16:10,280 --> 00:16:11,800
Yeah, they don't have long-term memory.

257
00:16:11,800 --> 00:16:12,920
That's the issue, kind of.

258
00:16:13,560 --> 00:16:16,680
You have different techniques and different ways

259
00:16:16,680 --> 00:16:21,160
to provide that and to enhance the context.

260
00:16:21,160 --> 00:16:24,840
But definitely, LLMs are just a black box

261
00:16:24,840 --> 00:16:26,920
that have a general knowledge about everything,

262
00:16:26,920 --> 00:16:28,680
but about nothing specific.

263
00:16:28,680 --> 00:16:29,800
Yeah, yeah, yeah.

264
00:16:29,800 --> 00:16:34,680
How you provide that and enhance that context is a whole world.

265
00:16:34,680 --> 00:16:35,560
Yeah, absolutely.

266
00:16:35,560 --> 00:16:36,040
Absolutely.

267
00:16:36,040 --> 00:16:38,920
It's a whole piece of art.

268
00:16:38,920 --> 00:16:39,640
Yeah.

269
00:16:39,720 --> 00:16:42,280
I got the art and the art system.

270
00:16:42,280 --> 00:16:44,360
They are used as plastic contents.

271
00:16:44,360 --> 00:16:44,920
Yeah.

272
00:16:44,920 --> 00:16:47,400
A different sensation.

273
00:16:47,400 --> 00:16:49,400
Yeah, that was something that was discussed also

274
00:16:49,400 --> 00:16:50,440
the last couple of weeks a lot.

275
00:16:50,440 --> 00:16:54,280
It's like basically building up specialized drag databases

276
00:16:54,280 --> 00:16:56,600
and having them signed by the domain experts.

277
00:16:56,600 --> 00:16:57,320
Yeah.

278
00:16:57,320 --> 00:17:00,120
And then it's kind of very easy to figure out

279
00:17:00,120 --> 00:17:01,960
what to use and what is actually,

280
00:17:02,680 --> 00:17:06,360
like it's very easy for the LLMs to grab

281
00:17:06,360 --> 00:17:07,720
domain-specific knowledge.

282
00:17:07,720 --> 00:17:08,440
Yeah.

283
00:17:08,520 --> 00:17:10,440
And I think that's huge.

284
00:17:10,440 --> 00:17:11,000
That's huge.

285
00:17:11,000 --> 00:17:13,640
It's like a, you know, like it's,

286
00:17:14,760 --> 00:17:17,720
because I see it in the same way as like the,

287
00:17:18,840 --> 00:17:20,760
a Casper of sees the chess computers.

288
00:17:20,760 --> 00:17:22,840
You know, I think that the strongest,

289
00:17:22,840 --> 00:17:24,680
it's not that the chess computers took over

290
00:17:24,680 --> 00:17:26,040
and they played against each other

291
00:17:26,040 --> 00:17:27,960
and, you know, dominate the chess world now.

292
00:17:27,960 --> 00:17:30,920
That's not, that's, the strongest systems

293
00:17:30,920 --> 00:17:33,080
are hybrid systems, are human machine systems.

294
00:17:33,080 --> 00:17:33,480
Yeah.

295
00:17:33,480 --> 00:17:37,480
And in terms of like now having vibe coded intensely

296
00:17:37,480 --> 00:17:39,080
for like six weeks straight, basically,

297
00:17:41,240 --> 00:17:43,800
it's clear to me that for the foreseeable future,

298
00:17:43,800 --> 00:17:44,920
this is what we have to do.

299
00:17:44,920 --> 00:17:48,360
We, we need to have this human agent interaction

300
00:17:48,360 --> 00:17:49,640
that we need to have also,

301
00:17:50,600 --> 00:17:54,840
we need to infuse domain expert knowledge

302
00:17:55,400 --> 00:17:56,600
into those systems.

303
00:17:56,600 --> 00:17:56,920
Yeah.

304
00:17:56,920 --> 00:17:57,800
And I think it's not so,

305
00:17:57,800 --> 00:18:00,680
it's a very natural system to do that to make this happen.

306
00:18:00,680 --> 00:18:01,480
Definitely.

307
00:18:01,480 --> 00:18:04,440
I am spreading exactly that with the beating head project.

308
00:18:05,080 --> 00:18:05,480
Yeah.

309
00:18:05,560 --> 00:18:07,320
The beating head project is something that

310
00:18:08,600 --> 00:18:10,520
I was just kissing with Joe here in,

311
00:18:10,520 --> 00:18:12,040
in several engineering and also

312
00:18:12,760 --> 00:18:15,320
Alex Gleason was interested on the topic.

313
00:18:15,320 --> 00:18:18,360
And so we are working kind of together on that

314
00:18:19,400 --> 00:18:20,360
very synchronously.

315
00:18:20,360 --> 00:18:23,480
But basically the beating head is a

316
00:18:23,480 --> 00:18:27,320
rack database that is made it of an NCP tool.

317
00:18:27,320 --> 00:18:29,080
Like I do have a NCP server

318
00:18:29,080 --> 00:18:31,320
that provides you access to a rack database.

319
00:18:31,880 --> 00:18:33,640
And this is something very, very cool that

320
00:18:34,600 --> 00:18:36,440
I think that I haven't seen,

321
00:18:36,440 --> 00:18:39,000
I haven't saw that other people are

322
00:18:39,000 --> 00:18:39,880
spreading this solution.

323
00:18:40,520 --> 00:18:41,800
Because, you know,

324
00:18:41,800 --> 00:18:43,240
rack databases are normally

325
00:18:43,960 --> 00:18:46,120
tailored for a specific system.

326
00:18:47,480 --> 00:18:50,120
But it really opens the door for a kind of

327
00:18:50,120 --> 00:18:54,440
black-able context or the mind-spare context, you know,

328
00:18:54,440 --> 00:18:57,480
because you can, the beating head have

329
00:18:58,920 --> 00:19:03,080
all the nips and context about the specific kinds

330
00:19:03,080 --> 00:19:04,600
and how to implement them.

331
00:19:04,600 --> 00:19:06,520
A very comprehensive documentation on that

332
00:19:06,520 --> 00:19:11,000
that can be retrieved with natural language.

333
00:19:11,000 --> 00:19:11,480
Very cool.

334
00:19:11,480 --> 00:19:13,400
Like LLMs can do that very easily.

335
00:19:14,280 --> 00:19:16,600
And I made it in a way that

336
00:19:17,960 --> 00:19:20,040
you have embed databases.

337
00:19:20,040 --> 00:19:22,680
So these whole contexts,

338
00:19:24,120 --> 00:19:27,960
the embedding database is all a single file

339
00:19:27,960 --> 00:19:30,040
that you can, for example, upload to Blossom.

340
00:19:30,680 --> 00:19:32,360
And then you can do the same,

341
00:19:32,360 --> 00:19:35,640
create a list of the rack databases that you have there.

342
00:19:36,200 --> 00:19:38,280
And then suddenly you have a

343
00:19:38,280 --> 00:19:41,480
plug-able context and plug-able experts.

344
00:19:42,200 --> 00:19:45,560
Or domain-specific content

345
00:19:45,560 --> 00:19:48,360
that you can plug into your system

346
00:19:48,360 --> 00:19:51,160
and benefit of that by providing real context

347
00:19:51,160 --> 00:19:54,200
about the domain that you want to work with your L&M.

348
00:19:54,200 --> 00:19:55,320
Yeah, I think that's awesome.

349
00:19:55,320 --> 00:19:57,080
And the awesome part about it is also,

350
00:19:57,080 --> 00:19:57,960
like you said before,

351
00:19:58,040 --> 00:19:59,800
the very fireability is baked in.

352
00:19:59,800 --> 00:20:02,040
Like everything is self-signed, so to speak.

353
00:20:02,040 --> 00:20:04,360
I mean, it's pure not-staring mens and again,

354
00:20:04,360 --> 00:20:05,880
like blossom uploads and so on.

355
00:20:05,880 --> 00:20:08,360
And the result of that is that

356
00:20:09,800 --> 00:20:12,840
you don't ever need an API key for anything.

357
00:20:12,840 --> 00:20:15,880
And that's something that's such a killer feature

358
00:20:15,880 --> 00:20:17,960
from a developer perspective,

359
00:20:17,960 --> 00:20:19,560
where you can just plug all this together.

360
00:20:20,120 --> 00:20:23,480
And you don't need to sign up anywhere

361
00:20:23,480 --> 00:20:25,720
and then turn an API key and anything like that.

362
00:20:25,720 --> 00:20:27,400
I mean, we will still have to deal with spam

363
00:20:27,400 --> 00:20:29,000
and rate-limiting and all those kind of things.

364
00:20:29,000 --> 00:20:29,480
Oh yeah.

365
00:20:29,480 --> 00:20:31,160
But we can do that easily with Web of Trust

366
00:20:31,160 --> 00:20:32,200
and e-cash basically.

367
00:20:32,200 --> 00:20:33,240
Everyone will have to deal with that.

368
00:20:33,240 --> 00:20:34,440
Everyone will have to deal with that.

369
00:20:34,440 --> 00:20:38,280
And the default way is like signups and KYC

370
00:20:38,280 --> 00:20:39,320
and all those kind of things.

371
00:20:39,320 --> 00:20:41,880
And the nicer way is basically Web of Trust and payments.

372
00:20:41,880 --> 00:20:44,040
Web of Trust and Sats.

373
00:20:44,040 --> 00:20:48,440
I mean, the original proof-of-work vision of,

374
00:20:49,080 --> 00:20:49,960
you know, like it was,

375
00:20:52,040 --> 00:20:55,000
yeah, it was both California and Adam Back had the idea,

376
00:20:55,000 --> 00:20:55,960
basically, of like,

377
00:20:55,960 --> 00:20:57,080
it was basically rate-limiting.

378
00:20:57,080 --> 00:20:59,480
It was like postage stamps for email sent,

379
00:20:59,480 --> 00:21:00,040
those kind of things.

380
00:21:00,040 --> 00:21:00,440
Yeah.

381
00:21:00,440 --> 00:21:02,280
It was a spam-prevention mechanism.

382
00:21:02,280 --> 00:21:02,760
Yeah.

383
00:21:02,760 --> 00:21:06,360
And we just used the difficulty adjusted

384
00:21:06,360 --> 00:21:08,680
spam-prevention mechanism that is Sats.

385
00:21:08,680 --> 00:21:09,160
Yeah.

386
00:21:10,120 --> 00:21:11,960
Sats and social security planning.

387
00:21:11,960 --> 00:21:12,600
Yeah, exactly.

388
00:21:13,160 --> 00:21:15,240
So, yeah, all that's really cool.

389
00:21:15,240 --> 00:21:17,160
Like for me, like again, playing around with that,

390
00:21:17,160 --> 00:21:18,200
and I don't, you know,

391
00:21:18,200 --> 00:21:19,880
I don't see myself as a developer anymore,

392
00:21:19,880 --> 00:21:21,320
that the train is long gone.

393
00:21:21,320 --> 00:21:24,040
But I am vibing again, you know, so there's that.

394
00:21:24,040 --> 00:21:27,480
And it's just so cool to be able to do all that and use all that

395
00:21:27,480 --> 00:21:30,040
and not have to worry about the sign-ups and API keys

396
00:21:30,040 --> 00:21:31,000
and whatnot.

397
00:21:31,000 --> 00:21:32,120
I mean, if you just,

398
00:21:32,120 --> 00:21:34,120
if you ever entered the Google Developer Console

399
00:21:34,120 --> 00:21:36,680
and tried to, you know, create an OAuth tool,

400
00:21:36,680 --> 00:21:39,800
or whatever, it's like, you know, like five days are gone

401
00:21:39,800 --> 00:21:40,840
and it still doesn't work.

402
00:21:41,640 --> 00:21:43,240
And here it's just, it's basically plug-and-play.

403
00:21:43,240 --> 00:21:44,360
So that's really cool.

404
00:21:44,360 --> 00:21:46,280
Yeah, that's the kind of authentication,

405
00:21:47,480 --> 00:21:50,440
hell, that I was talking about with the A to A

406
00:21:51,400 --> 00:21:54,600
they kind of make this protocol free and really interpretable

407
00:21:54,600 --> 00:21:57,960
or that really, something that really works for,

408
00:21:57,960 --> 00:22:00,600
because they are going to use for in the same

409
00:22:00,600 --> 00:22:03,320
problems that they have now that is, ah, fucking hell.

410
00:22:03,320 --> 00:22:06,760
Someone needs to say who is a legit user and who isn't.

411
00:22:06,760 --> 00:22:09,480
Yeah, and so that's with the DID people as well.

412
00:22:09,480 --> 00:22:12,200
You know, someone has to be the certificate authority,

413
00:22:12,200 --> 00:22:13,960
for example, and all those kind of things.

414
00:22:13,960 --> 00:22:16,520
And so all these systems, like massively centralized

415
00:22:16,520 --> 00:22:19,960
and basically turn into massively authoritarian systems

416
00:22:19,960 --> 00:22:23,160
because, you know, there will be, I don't know,

417
00:22:23,160 --> 00:22:25,000
like, you know, depending on whatever,

418
00:22:26,440 --> 00:22:28,200
depending on the political weather,

419
00:22:28,200 --> 00:22:31,080
it's like, okay, the developers are from Iran or whatever.

420
00:22:31,080 --> 00:22:32,680
Okay, we have to deplatform all of them.

421
00:22:32,680 --> 00:22:36,200
Like, if your IP is from this country, then, you know,

422
00:22:36,200 --> 00:22:38,280
like someone with a very large stick will come

423
00:22:38,280 --> 00:22:40,120
that will tell everyone, okay, you have to

424
00:22:40,120 --> 00:22:42,760
blacklist all those IPs, you have to remove all of that

425
00:22:42,760 --> 00:22:43,400
and so on.

426
00:22:43,400 --> 00:22:47,160
For, it doesn't, you know, like, and I mean,

427
00:22:47,160 --> 00:22:49,960
we've seen that very recently that a lot of the reasons

428
00:22:49,960 --> 00:22:53,480
are incredibly stupid for this kind of, you know,

429
00:22:53,480 --> 00:22:55,560
like measures and deplatforming and so on.

430
00:22:55,560 --> 00:22:59,400
And it's basically, in my opinion, it's an impossible problem.

431
00:22:59,400 --> 00:23:02,520
And there are no solutions to this problem in the sense that

432
00:23:02,520 --> 00:23:08,120
you can't, like, you either are okay with a central authoritarian authority.

433
00:23:09,480 --> 00:23:10,920
You know, certificate authority.

434
00:23:10,920 --> 00:23:12,600
It has authoritarian in its names.

435
00:23:13,560 --> 00:23:17,160
Or you have to be okay with the decentralized, messy,

436
00:23:17,880 --> 00:23:21,320
you know, doing it on the edges way of how Nostra is thinking about it,

437
00:23:21,320 --> 00:23:23,880
which means also, you know, like, it's not a perfect solution.

438
00:23:23,880 --> 00:23:26,280
It's like, Web of Trust is gameable.

439
00:23:26,280 --> 00:23:28,200
It's like, if someone has a lot of money,

440
00:23:28,200 --> 00:23:30,600
your rate limiting mechanism can be overwhelmed,

441
00:23:30,600 --> 00:23:31,320
all those kind of things.

442
00:23:31,320 --> 00:23:33,320
Like, I can pay to spam you and so on.

443
00:23:33,320 --> 00:23:36,440
But as we, as we said, like, these,

444
00:23:36,440 --> 00:23:39,160
these problems are problems that everyone have to pass through.

445
00:23:39,160 --> 00:23:39,800
Yeah.

446
00:23:39,800 --> 00:23:43,640
And there is no call them, like, a central bullet.

447
00:23:43,640 --> 00:23:44,440
Yeah, exactly.

448
00:23:45,560 --> 00:23:49,480
The trade-offs for these problems are so big and so

449
00:23:49,480 --> 00:23:53,560
limiting that they are going to fall in the, in, yeah.

450
00:23:53,560 --> 00:23:58,120
In something unusable, as right now is unusable,

451
00:23:58,120 --> 00:24:04,520
like all this, all of this new logging methods and so on.

452
00:24:05,080 --> 00:24:08,360
Like, it's making the life of very much difficulty.

453
00:24:08,360 --> 00:24:08,840
Yeah.

454
00:24:08,840 --> 00:24:09,320
Yeah.

455
00:24:09,320 --> 00:24:12,760
Oh, oh, oh, this is a good use case as well for kind of like,

456
00:24:12,760 --> 00:24:14,360
it had the best intentions.

457
00:24:14,360 --> 00:24:16,360
And I think it was just a little bit too complicated

458
00:24:16,920 --> 00:24:18,440
on the, on the technical side.

459
00:24:18,440 --> 00:24:21,320
And it also, like, it, it centralized massively

460
00:24:21,320 --> 00:24:24,040
because it's basically one school or Facebook becomes an old provider.

461
00:24:24,040 --> 00:24:26,680
No one else is going to care about anything else.

462
00:24:26,680 --> 00:24:27,480
I mean, yeah, in some sense.

463
00:24:27,480 --> 00:24:30,680
It was really funny how that, how that, you know,

464
00:24:30,680 --> 00:24:31,800
evolved in some sense.

465
00:24:32,680 --> 00:24:36,600
But yeah, we'll see how it goes with, with, I mean,

466
00:24:36,600 --> 00:24:40,120
there's still a long way until we can log in with not sure

467
00:24:40,120 --> 00:24:40,680
everywhere.

468
00:24:40,680 --> 00:24:44,520
But at least it has the potential to make it work.

469
00:24:44,920 --> 00:24:45,480
Yeah.

470
00:24:45,480 --> 00:24:48,840
So I mean, for me, one of the most exciting part and

471
00:24:50,360 --> 00:24:57,480
the reason why I'm spending time and energy in these fields of open

472
00:24:57,480 --> 00:25:01,800
source, sorry, I am jumping beyond tools and a lot of this.

473
00:25:02,840 --> 00:25:03,400
Sorry.

474
00:25:03,400 --> 00:25:05,400
Sorry.

475
00:25:09,640 --> 00:25:12,600
You know, if I'm spending all this energy and time,

476
00:25:13,160 --> 00:25:18,760
it's because there is a very clear path for open source in this,

477
00:25:18,760 --> 00:25:21,400
in this new technologies, like all the I.A.

478
00:25:22,360 --> 00:25:26,200
War is based on open source in the 50s, 70s.

479
00:25:26,200 --> 00:25:26,760
Yeah.

480
00:25:26,760 --> 00:25:30,280
All the work that have been done in that field is open source.

481
00:25:30,840 --> 00:25:35,000
Like all the technologies that we have, mostly all almost all of them

482
00:25:35,000 --> 00:25:36,200
is open source.

483
00:25:36,200 --> 00:25:38,680
So they are open source alternatives that was the original one.

484
00:25:39,640 --> 00:25:42,760
Do you, do you use open source models exclusively right now?

485
00:25:42,760 --> 00:25:44,840
Or yeah, you're a purist.

486
00:25:46,040 --> 00:25:48,040
I don't touch any open.

487
00:25:50,040 --> 00:25:53,240
I like to use my own models or proxies.

488
00:25:53,240 --> 00:25:54,840
So, so what have you used so far?

489
00:25:55,720 --> 00:25:57,160
I like models.

490
00:25:57,160 --> 00:25:58,200
Yeah, models.

491
00:25:58,200 --> 00:25:59,400
I use a lot of models.

492
00:25:59,400 --> 00:26:03,640
Like I, I like to spend a minute on to play with them locally.

493
00:26:03,640 --> 00:26:04,760
The ones that I can afford.

494
00:26:05,720 --> 00:26:08,280
But so you haven't upgraded your hardware yet.

495
00:26:09,400 --> 00:26:10,920
So it's tiny llama.

496
00:26:10,920 --> 00:26:12,280
Tiny llama.

497
00:26:12,280 --> 00:26:16,280
I got seven billion parameters, models, things like that.

498
00:26:16,280 --> 00:26:16,760
Okay.

499
00:26:16,760 --> 00:26:17,800
I can afford much more.

500
00:26:18,520 --> 00:26:23,320
But it's funny that most of the tiny ones are kind of good enough for a lot of things.

501
00:26:23,320 --> 00:26:23,800
Oh, yeah.

502
00:26:23,880 --> 00:26:24,600
It's really you.

503
00:26:24,600 --> 00:26:26,600
No, I think I need tiny models.

504
00:26:26,600 --> 00:26:34,840
It's also something that really excites me of this field because they give you some capacity,

505
00:26:34,840 --> 00:26:37,160
some computation.

506
00:26:37,160 --> 00:26:42,520
But it's not like an intelligent model that cannot solve big problems for you.

507
00:26:42,520 --> 00:26:45,480
But definitely they can do some stuff.

508
00:26:45,480 --> 00:26:50,040
And also you can train them and make them accurate for your specific uses.

509
00:26:50,040 --> 00:26:51,560
Yeah, for that specific task and so on.

510
00:26:51,640 --> 00:26:57,160
And I think that there, there's something really interesting in tiny models because they are going

511
00:26:57,160 --> 00:27:04,120
to allow like a things appear like this, this protocol by Stanford that is called Minions.

512
00:27:05,080 --> 00:27:07,320
I really like learning about the opinions.

513
00:27:07,320 --> 00:27:08,520
Minions model?

514
00:27:08,520 --> 00:27:09,400
Is it like a?

515
00:27:09,400 --> 00:27:10,280
No, it's not a model.

516
00:27:10,280 --> 00:27:11,400
It's a protocol itself.

517
00:27:12,200 --> 00:27:17,960
But what's presented by Stanford like a, does it have anything to do with the little yellow

518
00:27:18,040 --> 00:27:19,000
minions? Is that the idea?

519
00:27:19,000 --> 00:27:19,720
Yeah, yeah, yeah.

520
00:27:19,720 --> 00:27:22,680
It's a swarm of small, small, small, small models.

521
00:27:22,680 --> 00:27:23,160
Exactly.

522
00:27:23,160 --> 00:27:23,480
Exactly.

523
00:27:23,480 --> 00:27:23,800
Great.

524
00:27:23,800 --> 00:27:24,200
It's an internet.

525
00:27:25,240 --> 00:27:27,000
It's a cyclic, it's a cyclic that.

526
00:27:27,000 --> 00:27:27,480
Okay.

527
00:27:27,480 --> 00:27:34,760
So the protocol describes a way to have this swarm of minions to something useful.

528
00:27:34,760 --> 00:27:35,640
Yeah, exactly.

529
00:27:35,640 --> 00:27:37,480
Even though every single one is super stupid.

530
00:27:37,480 --> 00:27:37,960
Yeah.

531
00:27:37,960 --> 00:27:38,040
Okay.

532
00:27:38,040 --> 00:27:47,480
So the idea basically is you can kind of decompose a task into different minions that you can have.

533
00:27:47,960 --> 00:27:53,480
And these minions are going to be collaborating with a standard provider that can be a more powerful

534
00:27:53,480 --> 00:27:54,760
model or a proprietary model.

535
00:27:55,640 --> 00:28:01,480
But the key here is that the minions are going to be anonymizing, anonymizing the data

536
00:28:02,040 --> 00:28:07,080
and make it more specific, creating more specific tasks for the big model.

537
00:28:07,720 --> 00:28:10,120
So you are going to be anonymizing the data.

538
00:28:10,120 --> 00:28:11,640
So you still have, you still have a big model.

539
00:28:11,640 --> 00:28:13,480
You still have a generalist to speak.

540
00:28:13,480 --> 00:28:17,400
You still have a big model architect that has the overview of those kinds of things.

541
00:28:17,400 --> 00:28:20,040
Yeah, I mean, in orchestrator for the minions in some sense.

542
00:28:20,040 --> 00:28:20,280
Yeah.

543
00:28:20,280 --> 00:28:22,200
Like tiny models are stupid.

544
00:28:22,200 --> 00:28:22,840
Yeah, absolutely.

545
00:28:22,840 --> 00:28:23,320
Pretty dumb.

546
00:28:24,280 --> 00:28:29,400
But they are really good for a specific, very concise task.

547
00:28:30,280 --> 00:28:30,760
So they can...

548
00:28:30,760 --> 00:28:31,240
Give me one second.

549
00:28:31,240 --> 00:28:32,760
I'm not sure if I hit record, actually.

550
00:28:32,760 --> 00:28:34,120
It's like, yeah, there we go.

551
00:28:34,120 --> 00:28:35,080
That's interesting.

552
00:28:35,720 --> 00:28:36,680
It could be better than I'm adding.

553
00:28:36,680 --> 00:28:37,720
No, no, no, no, no.

554
00:28:37,720 --> 00:28:38,360
It's still good.

555
00:28:38,360 --> 00:28:39,000
It's still good.

556
00:28:39,000 --> 00:28:41,720
I always have a backup kind of, but it's like, I don't know.

557
00:28:41,720 --> 00:28:45,640
You know, now that I have had half of my coffee, the brain starts firing again.

558
00:28:45,640 --> 00:28:46,680
I think it's something to think about.

559
00:28:46,680 --> 00:28:47,640
But it's...

560
00:28:47,640 --> 00:28:52,680
Because what is interesting, in nature, you have a lot of stupid things

561
00:28:52,680 --> 00:28:57,000
without the big orchestrator, and it still works and produces something smart,

562
00:28:57,000 --> 00:28:58,520
like an ant colony or what have you.

563
00:28:58,520 --> 00:28:59,160
No.

564
00:28:59,160 --> 00:29:01,800
But in, yeah, yeah.

565
00:29:01,800 --> 00:29:06,040
So that's why I was asking whether there's still a big one in some sense in charge,

566
00:29:06,040 --> 00:29:09,400
because I talked about this, I think we've gathered as well,

567
00:29:09,400 --> 00:29:14,360
but with Pablo in particular that, to me,

568
00:29:14,360 --> 00:29:23,080
it's fascinating that we see the split between generalists and specialists,

569
00:29:23,080 --> 00:29:24,040
and it has to happen.

570
00:29:24,040 --> 00:29:25,480
Like, it has to be that way.

571
00:29:25,480 --> 00:29:28,040
All the cognitive scientists will tell you it has to be this way,

572
00:29:28,040 --> 00:29:31,320
because you cannot specialize and generalize at the same time.

573
00:29:31,320 --> 00:29:32,920
It's in contradiction, basically.

574
00:29:35,320 --> 00:29:38,760
So what I think we could have also, in theory, in some sense,

575
00:29:38,760 --> 00:29:41,320
is something like you're describing,

576
00:29:41,320 --> 00:29:45,480
but without a smart orchestrator model in charge in some sense.

577
00:29:45,480 --> 00:29:46,280
I mean, I mean...

578
00:29:46,280 --> 00:29:47,160
Well, I'll have to look into that.

579
00:29:47,160 --> 00:29:48,040
That sounds fascinating.

580
00:29:48,040 --> 00:29:50,280
No, it's a very, very interesting project, because...

581
00:29:50,280 --> 00:29:51,640
You said Stanford?

582
00:29:51,640 --> 00:29:52,440
Stanford, yeah.

583
00:29:52,440 --> 00:29:53,640
University.

584
00:29:53,640 --> 00:29:58,280
But, say, it's very interesting because it has blurred different topics

585
00:29:58,280 --> 00:30:01,480
that we are going to need to explore, like privacy, in IA,

586
00:30:01,480 --> 00:30:08,360
or in the interaction with LLMs, and how to leverage a local compute

587
00:30:09,000 --> 00:30:11,400
that you can afford with these tiny models.

588
00:30:11,400 --> 00:30:13,640
Ah, actually, yeah, yeah, yeah.

589
00:30:13,640 --> 00:30:15,480
So that not everything concentrates in, like,

590
00:30:15,480 --> 00:30:17,080
Facebook data centers, basically.

591
00:30:17,080 --> 00:30:18,280
Yeah, yeah, exactly.

592
00:30:18,280 --> 00:30:22,680
So the key here is that you can get your queries or your prompts,

593
00:30:22,680 --> 00:30:23,880
anonymized.

594
00:30:23,880 --> 00:30:25,400
Yeah, I see.

595
00:30:25,400 --> 00:30:29,640
So, like, SETI at home, like SETI at home.

596
00:30:29,640 --> 00:30:31,960
You know that, from the old days,

597
00:30:31,960 --> 00:30:35,320
searching for extraterrestrial life at home,

598
00:30:35,320 --> 00:30:37,080
of, like, running a screensafe and stuff.

599
00:30:37,080 --> 00:30:38,520
So that would be minions at home,

600
00:30:38,520 --> 00:30:43,240
where it's just, like, you have whatever compute packets that you run,

601
00:30:43,240 --> 00:30:48,120
and, yeah, well, yeah, but they will run into the issue of payments.

602
00:30:48,120 --> 00:30:53,000
Because you want to sell your idle GPU, or CPU, or RAM, or what have you.

603
00:30:53,000 --> 00:30:57,080
And, I mean, that's what, actually, in the first cohort,

604
00:30:57,080 --> 00:30:59,480
we talked a little bit about that with blinded compute,

605
00:30:59,480 --> 00:31:01,320
and I talked with Kelly about that as well,

606
00:31:01,320 --> 00:31:03,720
like blinded compute and serial knowledge proofs,

607
00:31:03,720 --> 00:31:04,920
and those kind of things.

608
00:31:05,080 --> 00:31:07,400
That's going to be so interesting when, like,

609
00:31:08,280 --> 00:31:11,160
it's not an if, but it's a when, like, when we get this to work.

610
00:31:12,600 --> 00:31:15,240
Yeah, I think that, as I was saying at the beginning,

611
00:31:15,240 --> 00:31:20,760
we are in this, like, the left side of this publicity cult,

612
00:31:20,760 --> 00:31:21,720
like at the hype.

613
00:31:21,720 --> 00:31:23,560
Yeah, we're also at the left side of the bell curve.

614
00:31:23,560 --> 00:31:24,360
Yeah.

615
00:31:24,360 --> 00:31:26,440
I hope this is a good part.

616
00:31:28,120 --> 00:31:33,720
But for me, the fascinating part is that no sir kind of fits well

617
00:31:33,720 --> 00:31:39,960
in all of this new party, because it has some kind of answers

618
00:31:39,960 --> 00:31:43,240
to the program, to the programatics that we are starting to see.

619
00:31:44,280 --> 00:31:48,360
And for me, that is fascinating because it's kind of getting right.

620
00:31:48,360 --> 00:31:52,440
Yeah, I mean, it's also, it's not super surprising.

621
00:31:52,440 --> 00:31:54,680
I mean, no sir was built, Bitcoin was built for people,

622
00:31:55,320 --> 00:31:56,760
for people at the edges.

623
00:31:56,760 --> 00:31:57,880
It's like, what do you do?

624
00:31:57,880 --> 00:32:00,200
What do you use when you don't have an identity?

625
00:32:00,200 --> 00:32:01,960
Like, you're from Ghana, you're 14,

626
00:32:02,040 --> 00:32:05,720
but you have access to some compute, but you don't have a passport,

627
00:32:05,720 --> 00:32:09,400
you don't have an ID, you don't even know when your birthday is,

628
00:32:09,400 --> 00:32:10,600
those kind of things, you know.

629
00:32:10,600 --> 00:32:13,240
And agents have the same problem.

630
00:32:13,240 --> 00:32:17,160
They are not, like, in the eyes of the snake, they are not people.

631
00:32:17,160 --> 00:32:20,840
And so they can't KYC, you know, like, they have all these issues and so on.

632
00:32:20,840 --> 00:32:23,160
And that's on the identity side.

633
00:32:23,160 --> 00:32:25,800
And that's why, I mean, most people thought a lot,

634
00:32:25,800 --> 00:32:28,760
and Bitcoin people in general, they thought long and hard about identity.

635
00:32:28,760 --> 00:32:29,640
What do you need?

636
00:32:29,640 --> 00:32:33,560
What kind of identity do you actually need to authenticate yourself

637
00:32:33,560 --> 00:32:36,360
towards a compute system, in some sense?

638
00:32:36,360 --> 00:32:39,080
And it's not a passport, like, that's, I mean,

639
00:32:39,080 --> 00:32:41,240
cypherpunk's figured in the South a long time ago,

640
00:32:41,240 --> 00:32:44,600
it must not be a state-sanctioned identity, because that, again,

641
00:32:44,600 --> 00:32:48,200
if you have a certificate authority, you run into authoritarian problems.

642
00:32:48,200 --> 00:32:49,800
That's just how it works.

643
00:32:49,800 --> 00:32:52,600
And the other part on an ostracile, which I mean, you know,

644
00:32:55,400 --> 00:32:59,160
Bitcoin is in particular kind of thought long and hard about,

645
00:32:59,160 --> 00:33:01,880
Nixabo in particular, is the micro-payment side.

646
00:33:01,880 --> 00:33:05,240
And the kind of, what is the difference between credit and money?

647
00:33:05,240 --> 00:33:09,880
And why do you, why can't you use credit, basically, for most of those things?

648
00:33:09,880 --> 00:33:13,480
And you always run into these limits, settlement limits,

649
00:33:13,480 --> 00:33:18,440
with micro-payments and also cognitive decision limits and all those kind of things.

650
00:33:18,440 --> 00:33:23,080
And I don't know of another group of people that thought about this problem

651
00:33:23,640 --> 00:33:25,640
for an extended period of time.

652
00:33:25,640 --> 00:33:27,640
It actually actively worked on solutions.

653
00:33:27,640 --> 00:33:31,400
And so, again, it's not surprising.

654
00:33:31,400 --> 00:33:34,440
It's very poetic that noster is kind of the perfect fit,

655
00:33:34,440 --> 00:33:37,640
but it's also not very surprising, because, again,

656
00:33:37,640 --> 00:33:40,040
agents, they're going to need some sort of identity,

657
00:33:40,040 --> 00:33:42,200
and it's not going to be state-sanctioned identity.

658
00:33:42,200 --> 00:33:46,760
And they need some sort of payments, and it's not going to be stripe accounts.

659
00:33:46,760 --> 00:33:48,360
You know, it's not going to be credit rates.

660
00:33:48,360 --> 00:33:51,240
Like, they can't, they just can't do it.

661
00:33:51,240 --> 00:33:56,040
And we already have agents that can autonomously sap with NIP-60.

662
00:33:56,760 --> 00:34:01,400
And we also have, I mean, you build stuff where you can pay agents,

663
00:34:01,400 --> 00:34:03,800
and agents can pay lightning and voices, and so on.

664
00:34:03,800 --> 00:34:05,400
So all this stuff already works.

665
00:34:05,400 --> 00:34:06,600
It's just like no one knows about it.

666
00:34:06,600 --> 00:34:11,400
There's like three people that are experimenting with that right now.

667
00:34:11,400 --> 00:34:13,240
Yeah, and also we have a little bit early.

668
00:34:13,240 --> 00:34:14,600
Our own problematics, you know.

669
00:34:14,600 --> 00:34:15,080
Oh, sure.

670
00:34:15,640 --> 00:34:19,640
About all the things and the path and specifications and so on.

671
00:34:19,640 --> 00:34:21,080
But there is a different distribution.

672
00:34:21,080 --> 00:34:23,480
I mean, none of that is perfect by any means,

673
00:34:23,480 --> 00:34:30,680
but the thing is that it is the right solution in theory, so to speak.

674
00:34:30,680 --> 00:34:32,920
And it also already works in practice.

675
00:34:32,920 --> 00:34:36,600
And I think that's why I urge everyone to kind of, you know,

676
00:34:36,600 --> 00:34:37,400
write a better tutorial.

677
00:34:37,400 --> 00:34:41,640
A lot of people play around with it, because the stuff is further along than you would think.

678
00:34:42,200 --> 00:34:44,280
And no one really knows about it yet,

679
00:34:44,280 --> 00:34:45,960
especially on the AI side.

680
00:34:45,960 --> 00:34:48,120
The AI people don't really know about this stuff yet.

681
00:34:48,120 --> 00:34:51,320
Even in our little noster bubble, most people don't know

682
00:34:52,280 --> 00:34:55,640
that we actually have autonomous agents already that can set things.

683
00:34:55,640 --> 00:34:57,880
Yeah, I'm fucking hell, I'm fucking hell.

684
00:35:00,280 --> 00:35:05,400
And that's true, because, you know, it's going to be permissionless or hell.

685
00:35:05,400 --> 00:35:07,400
Like, it's not a middle ground.

686
00:35:07,400 --> 00:35:11,560
Like, it cannot be semi-permissioned and still working.

687
00:35:11,560 --> 00:35:13,240
It has to be totally permissionless.

688
00:35:13,240 --> 00:35:17,320
Like, I should be able to do anything within a local scope.

689
00:35:17,320 --> 00:35:19,560
I don't need the global scope for the majority of things.

690
00:35:19,560 --> 00:35:23,320
Like, for example, if I want to automate my house

691
00:35:23,320 --> 00:35:28,280
and have some NCP tools that control the lights on my house,

692
00:35:28,280 --> 00:35:32,520
so I don't want to ask permission to win or to add.

693
00:35:32,520 --> 00:35:37,000
You also don't want to send all that stuff to the cloud, you know,

694
00:35:37,000 --> 00:35:38,520
and all those kind of things.

695
00:35:40,920 --> 00:35:46,600
Yeah, that's the cool thing, that also the noster architecture and EKASH as well.

696
00:35:47,160 --> 00:35:51,960
And it's very usable in a local-first environment.

697
00:35:51,960 --> 00:35:54,200
It's just like, you can set up your home relay,

698
00:35:54,200 --> 00:35:56,280
you can, you know, do all those kind of things,

699
00:35:56,280 --> 00:35:59,000
you can set up a minted home, like who cares.

700
00:35:59,000 --> 00:36:03,160
And you just, you know, all the data stays in your house, basically.

701
00:36:03,160 --> 00:36:05,000
All the stuff stays in your house, you don't, again,

702
00:36:05,000 --> 00:36:07,960
you don't need an API key, you don't need the cloud for anything.

703
00:36:07,960 --> 00:36:10,360
And it's still super powerful stuff.

704
00:36:10,360 --> 00:36:13,400
So, yeah, we'll see.

705
00:36:13,400 --> 00:36:16,280
I mean, again, like you said, none of that works.

706
00:36:16,280 --> 00:36:20,520
We have our own issues, working through all of that.

707
00:36:20,520 --> 00:36:21,880
Yeah, we are discovering that, man.

708
00:36:23,640 --> 00:36:25,240
And definitely, there are going to be solutions,

709
00:36:25,240 --> 00:36:30,120
and ways to, more clear ways to do things once we discover

710
00:36:30,120 --> 00:36:32,360
what is the real potential for all this,

711
00:36:32,360 --> 00:36:34,680
because now everyone is experimenting.

712
00:36:34,680 --> 00:36:38,120
And that's why it's a very exciting technology, you know,

713
00:36:38,120 --> 00:36:41,880
like a technology that doesn't fully work.

714
00:36:41,960 --> 00:36:48,680
It works sometimes, but it's not yet like a solid, in a way.

715
00:36:48,680 --> 00:36:51,800
So that makes that this technology exciting.

716
00:36:51,800 --> 00:36:54,520
Yeah, I hope that more people are going to pick it up.

717
00:36:54,520 --> 00:36:56,200
Let's do one more ring around here,

718
00:36:56,200 --> 00:36:57,960
and then let's go to the park or something.

719
00:36:57,960 --> 00:37:02,440
But I hope that more people are going to play around with DB MCP,

720
00:37:02,440 --> 00:37:03,640
and all those kinds of things,

721
00:37:03,640 --> 00:37:05,640
and just actually publish code snippets,

722
00:37:05,640 --> 00:37:07,880
and publish MCP tools, and so on.

723
00:37:07,880 --> 00:37:09,560
Because it's super easy to do, actually,

724
00:37:09,560 --> 00:37:11,320
and we have the marketplace already.

725
00:37:11,320 --> 00:37:12,440
And once you have a marketplace,

726
00:37:12,440 --> 00:37:15,720
I think that's what I talked about this briefly,

727
00:37:15,720 --> 00:37:17,720
I think on Friday night or something.

728
00:37:18,840 --> 00:37:21,960
I think most awesome software,

729
00:37:21,960 --> 00:37:23,880
most great things that we have today,

730
00:37:24,760 --> 00:37:27,720
came from the marketplace approach in some sense.

731
00:37:27,720 --> 00:37:30,120
And the way I mean it is like most browsers,

732
00:37:30,120 --> 00:37:32,440
for example, that we have that are fantastic.

733
00:37:33,080 --> 00:37:35,240
They were made fantastic by a community of people

734
00:37:35,240 --> 00:37:37,000
that was writing extensions.

735
00:37:38,200 --> 00:37:40,200
And then the extensions got merged upstream,

736
00:37:40,200 --> 00:37:42,360
and it just became a normal thing, and those kinds of things.

737
00:37:42,360 --> 00:37:47,640
And we really need to tap into that, basically,

738
00:37:47,640 --> 00:37:50,440
into the wisdom of the crowd, in some sense.

739
00:37:50,440 --> 00:37:51,160
Games as well.

740
00:37:52,280 --> 00:37:54,280
The biggest games that we have now in the world,

741
00:37:54,280 --> 00:37:56,280
like computer games, they were all mods.

742
00:37:56,280 --> 00:37:57,240
They were all game mods.

743
00:37:57,240 --> 00:37:59,080
Like Dota came from Walk of Three,

744
00:37:59,080 --> 00:38:03,720
basically it was just a defense of the Asians mod.

745
00:38:03,720 --> 00:38:05,480
And Counter Strike came from Half-Life,

746
00:38:05,480 --> 00:38:07,640
that's basically, and all those kinds of things.

747
00:38:07,640 --> 00:38:13,320
And I can't wait to see what other people are going to build,

748
00:38:13,320 --> 00:38:15,000
what kind of MCP tools they are going to build,

749
00:38:15,000 --> 00:38:16,280
what kind of DVM's they build.

750
00:38:17,400 --> 00:38:20,040
Once they realize that there is this marketplace,

751
00:38:20,040 --> 00:38:22,040
and also an important part, I think,

752
00:38:22,040 --> 00:38:24,520
will also be like, you can immediately make money with it.

753
00:38:24,520 --> 00:38:27,400
I mean, right now, it's like you're going to make 20 stats.

754
00:38:27,400 --> 00:38:29,400
I don't know, 21 probably.

755
00:38:30,680 --> 00:38:35,320
But the fact that money and identity and reputation

756
00:38:35,320 --> 00:38:37,960
is infused into all of that, it makes it so exciting.

757
00:38:37,960 --> 00:38:39,160
Because you can actually,

758
00:38:39,160 --> 00:38:40,360
it's like a permissionless App Store,

759
00:38:40,360 --> 00:38:42,200
and you can publish to it.

760
00:38:42,200 --> 00:38:44,120
And App Store shows, for example,

761
00:38:44,120 --> 00:38:45,880
that this is really already working.

762
00:38:45,880 --> 00:38:47,800
And App Store is very, very young as well.

763
00:38:47,800 --> 00:38:49,800
I mean, it came out of the first cohort, basically.

764
00:38:49,800 --> 00:38:51,000
So it's less than a year old.

765
00:38:54,120 --> 00:38:56,600
That's why I have so high hopes for DVMTP as well,

766
00:38:56,600 --> 00:38:57,480
and all those kind of things.

767
00:38:57,480 --> 00:39:00,120
I think I also think, I totally agree with you.

768
00:39:00,120 --> 00:39:03,800
And I also think that it's such a powerful thing,

769
00:39:03,880 --> 00:39:09,400
because all this kind of fake sensation of control,

770
00:39:09,400 --> 00:39:13,640
that these big companies have about us,

771
00:39:13,640 --> 00:39:14,520
it's not going to work.

772
00:39:14,520 --> 00:39:17,960
Because the fucking viruses that are in the App Store are in the place for.

773
00:39:19,160 --> 00:39:20,040
But a part of that,

774
00:39:20,040 --> 00:39:25,400
I think that we have this sensation of control by them,

775
00:39:25,400 --> 00:39:28,040
because we don't have the right way to collaborate.

776
00:39:29,880 --> 00:39:32,840
We were talking the other day about the whole YouTube.

777
00:39:32,840 --> 00:39:36,680
It became a fucking hell to use to enjoy watching videos and so on,

778
00:39:36,680 --> 00:39:41,240
because you are going to get limited or you are going to get a...

779
00:39:41,240 --> 00:39:43,400
If you use a VPN, you are not welcome anymore.

780
00:39:45,560 --> 00:39:51,240
I literally, I wanted to download a video that I myself uploaded,

781
00:39:51,240 --> 00:39:53,880
that I myself recorded, of a talk that I myself gave,

782
00:39:54,440 --> 00:39:55,800
from my own YouTube account.

783
00:39:55,800 --> 00:39:56,120
Yeah.

784
00:39:56,120 --> 00:39:56,760
And I couldn't.

785
00:39:56,760 --> 00:39:57,720
I just, I couldn't.

786
00:39:57,720 --> 00:39:59,080
It wouldn't allow me to record.

787
00:39:59,080 --> 00:40:03,160
I tried for way, for longer than I would like to admit.

788
00:40:03,160 --> 00:40:05,960
But for me, disabling my VPN is very involved,

789
00:40:05,960 --> 00:40:07,560
because I basically, like,

790
00:40:07,560 --> 00:40:10,840
I have two routers, and on every router I run a separate VPN.

791
00:40:10,840 --> 00:40:13,160
And so just, I'm not going to go into my router settings

792
00:40:13,160 --> 00:40:16,040
and disable all VPNs, and just, I just gave up.

793
00:40:16,040 --> 00:40:17,000
And I was like,

794
00:40:17,000 --> 00:40:19,800
okay, I guess I'm never going to upload anything to YouTube anymore.

795
00:40:20,360 --> 00:40:24,440
Yeah, but I think that's, that's the, and we discussed this also a little bit

796
00:40:24,440 --> 00:40:26,040
the last couple of weeks.

797
00:40:26,920 --> 00:40:30,520
I hope that Noster can provide a kind of safe haven for that as well,

798
00:40:30,520 --> 00:40:32,680
where it's just like, you have,

799
00:40:33,320 --> 00:40:35,080
and the DVMs will play a part in that as well,

800
00:40:35,080 --> 00:40:36,520
because you don't want to do all that manually.

801
00:40:36,520 --> 00:40:38,360
But basically, I have a YouTube account,

802
00:40:38,360 --> 00:40:39,640
I have some videos on it.

803
00:40:39,640 --> 00:40:43,640
I want to, basically, grab them and archive them on my own,

804
00:40:43,640 --> 00:40:45,480
on my private blossom, for example.

805
00:40:45,480 --> 00:40:48,200
You know, and it's just like, I'm willing to pay for that.

806
00:40:48,200 --> 00:40:51,320
So if there's a DVM for that, if someone creates a DVM for that,

807
00:40:51,320 --> 00:40:53,880
please let me know, because like, you'll have my sets,

808
00:40:53,880 --> 00:40:55,400
so that I can have my videos again.

809
00:40:55,800 --> 00:40:57,000
That's what I was talking about.

810
00:40:57,720 --> 00:41:01,720
If we found the good ways to collaborate between us

811
00:41:01,720 --> 00:41:03,800
and share the resources that we have,

812
00:41:03,800 --> 00:41:07,240
for example, the IP, the domestic IP that we have,

813
00:41:07,800 --> 00:41:10,600
and everyone have, when they have internet,

814
00:41:10,600 --> 00:41:11,960
yeah, kind of VPN.

815
00:41:11,960 --> 00:41:14,920
Yeah, you are just listening to your IP service.

816
00:41:14,920 --> 00:41:17,320
If you have some hard time space left,

817
00:41:17,320 --> 00:41:19,880
and it's like, say for the GPUs and CPUs.

818
00:41:19,880 --> 00:41:20,520
Exactly.

819
00:41:20,520 --> 00:41:24,520
So I think that this kind of fake sensation of control,

820
00:41:24,520 --> 00:41:26,600
that these companies have about us,

821
00:41:26,600 --> 00:41:30,120
is because we haven't found yet the right way to collaborate.

822
00:41:30,120 --> 00:41:30,920
Yeah, yeah.

823
00:41:30,920 --> 00:41:33,800
And I think that no sir have something,

824
00:41:33,800 --> 00:41:37,640
got something right in the approach to communication,

825
00:41:38,120 --> 00:41:40,920
and the architecture of the network and so on,

826
00:41:40,920 --> 00:41:43,240
that we can find the right way to collaborate

827
00:41:43,240 --> 00:41:45,720
and really reduce the amount of control

828
00:41:45,720 --> 00:41:48,840
that these companies and these entities have over us.

829
00:41:49,480 --> 00:41:51,000
Yeah, let's hope that we'll figure it out.

830
00:41:51,000 --> 00:41:52,680
I mean, it's about time.

831
00:41:52,680 --> 00:41:57,480
It's like, it's stuff will only get more restrictive,

832
00:41:57,480 --> 00:41:59,320
I think, and more crazy kind of,

833
00:41:59,320 --> 00:42:01,160
and it's just now everyone.

834
00:42:01,160 --> 00:42:03,880
I mean, it's funny how it works now,

835
00:42:03,880 --> 00:42:08,440
because basically all these giant companies

836
00:42:08,440 --> 00:42:11,640
were built up to lock everyone in.

837
00:42:12,520 --> 00:42:15,880
It's like giant prisons in some sense,

838
00:42:15,880 --> 00:42:19,800
and you're locked in this golden cage,

839
00:42:19,800 --> 00:42:21,160
and you're getting advertisements

840
00:42:21,160 --> 00:42:22,120
jumped down your throat.

841
00:42:23,880 --> 00:42:24,920
It's fucking brutal.

842
00:42:24,920 --> 00:42:26,280
It's fucking brutal, yes.

843
00:42:26,280 --> 00:42:28,040
And that's how they made their wealth, basically.

844
00:42:28,040 --> 00:42:30,920
And now, it's like this switcher rule,

845
00:42:30,920 --> 00:42:32,920
where it's still a cage,

846
00:42:32,920 --> 00:42:36,040
but now at the same time that this is happening,

847
00:42:36,040 --> 00:42:39,160
you use this training data, basically.

848
00:42:39,160 --> 00:42:42,840
And so I think this will only get worse,

849
00:42:42,840 --> 00:42:45,560
and more political, and like, how,

850
00:42:45,560 --> 00:42:49,960
what is the LLM allowed to say, and see,

851
00:42:49,960 --> 00:42:51,240
and what do you train it on,

852
00:42:51,240 --> 00:42:53,480
and it's like, we see this with the Chinese selling them,

853
00:42:53,480 --> 00:42:56,520
so where it's like, you know, I don't know,

854
00:42:56,520 --> 00:42:57,720
Tian Man Square, or whatever,

855
00:42:57,720 --> 00:43:01,000
like whatever the usual phrases are to test these systems,

856
00:43:01,000 --> 00:43:03,000
and it's like, oh, I can't tell you anything about this,

857
00:43:03,000 --> 00:43:03,880
because it never happened.

858
00:43:05,960 --> 00:43:08,200
And I mean, and don't make the mistake that,

859
00:43:08,200 --> 00:43:09,720
you know, the glorious West,

860
00:43:09,720 --> 00:43:11,960
the glorious US has everything perfect,

861
00:43:11,960 --> 00:43:13,480
and doesn't meddle with history,

862
00:43:13,480 --> 00:43:14,760
and doesn't meddle with censorship.

863
00:43:14,760 --> 00:43:16,680
It's just a different kind of, you know,

864
00:43:16,680 --> 00:43:18,360
manipulation and censorship.

865
00:43:19,080 --> 00:43:23,000
So I really think that when we reach the top

866
00:43:23,000 --> 00:43:26,280
of the publicity club, that it's going to happen

867
00:43:26,280 --> 00:43:30,040
eventually in one year, two years, do you name it?

868
00:43:30,040 --> 00:43:30,920
I don't know.

869
00:43:30,920 --> 00:43:32,520
But when we start to go down,

870
00:43:32,520 --> 00:43:33,480
I think that it's going to,

871
00:43:33,480 --> 00:43:35,400
the going down is going to be driving

872
00:43:35,400 --> 00:43:39,000
by these fundamental questions about all of this,

873
00:43:39,000 --> 00:43:45,640
like from where the LLM get the data that...

874
00:43:45,640 --> 00:43:47,800
What is true, what is true?

875
00:43:47,800 --> 00:43:48,760
Yeah, what is true?

876
00:43:48,760 --> 00:43:49,640
It's going to be the question.

877
00:43:52,280 --> 00:43:55,640
All the intellectual property people

878
00:43:55,640 --> 00:43:58,840
are going to be questioning all these things,

879
00:43:58,840 --> 00:44:01,720
artists, they're going to be a lot of questions

880
00:44:01,720 --> 00:44:04,360
that are going to drive a lot of discussion.

881
00:44:04,360 --> 00:44:06,600
And I think that this discussion are going to

882
00:44:06,600 --> 00:44:11,480
slow down in some way, or at least make people realize

883
00:44:11,480 --> 00:44:14,600
what is the potential, or what is the real,

884
00:44:14,600 --> 00:44:17,720
it's cold, it's a big one.

885
00:44:17,720 --> 00:44:19,400
Because I think that we are not realizing yet

886
00:44:19,400 --> 00:44:21,560
because we are so excited and so hyped

887
00:44:21,560 --> 00:44:25,400
for these new technologies that we are not answering

888
00:44:25,400 --> 00:44:27,800
those questions yet, as well as we are not answering

889
00:44:27,800 --> 00:44:29,160
vibing into this topic.

890
00:44:29,160 --> 00:44:31,320
Yeah, right now.

891
00:44:31,320 --> 00:44:33,560
Yeah, I don't disagree, I don't disagree.

892
00:44:33,560 --> 00:44:36,440
We are going to realize that we are going to have

893
00:44:36,440 --> 00:44:39,640
to answer questions that we haven't answered yet.

894
00:44:39,640 --> 00:44:41,320
So I think that it's going to drive up

895
00:44:42,200 --> 00:44:43,720
at the realization.

896
00:44:43,720 --> 00:44:44,840
I'll have to undress.

897
00:44:44,840 --> 00:44:46,680
I feel the flames of hell already.

898
00:44:48,600 --> 00:44:49,480
Yeah, it's going to be hard.

899
00:44:49,480 --> 00:44:52,360
It's, I think, I think with,

900
00:44:53,480 --> 00:44:56,040
yeah, I mean, give me a second.

901
00:44:57,480 --> 00:45:00,920
So it turns out that, you know, the founding fathers

902
00:45:00,920 --> 00:45:03,720
of the US, oh my lord, is it still rolling?

903
00:45:04,600 --> 00:45:05,800
Is it bulletproof?

904
00:45:05,800 --> 00:45:06,440
No one knows.

905
00:45:07,080 --> 00:45:08,040
Oh yeah, it still works.

906
00:45:08,040 --> 00:45:08,440
Let's go.

907
00:45:08,440 --> 00:45:13,800
I'm just glad I didn't drop it into the sea.

908
00:45:13,800 --> 00:45:15,880
Like we would be entirely possible as well.

909
00:45:17,080 --> 00:45:18,600
But the, you know, I mean,

910
00:45:20,280 --> 00:45:21,880
I feel like we went through all that before.

911
00:45:22,440 --> 00:45:24,200
Okay, without LLMs, but still,

912
00:45:25,880 --> 00:45:28,440
probably without LLMs, who knows what kind of technology

913
00:45:28,440 --> 00:45:29,240
the Egyptians had.

914
00:45:30,600 --> 00:45:37,640
But I mean, there's a reason why we don't dictate truth.

915
00:45:38,600 --> 00:45:41,560
There's a reason why everyone should be hurt

916
00:45:41,560 --> 00:45:42,920
and everyone should be allowed to speak

917
00:45:42,920 --> 00:45:44,520
and why we landed on the First Amendment

918
00:45:44,520 --> 00:45:46,360
and why the logos is sacred and so on.

919
00:45:46,360 --> 00:45:48,040
I mean, I always repeat the same things.

920
00:45:48,040 --> 00:45:50,280
But it's, there is a deep seated reason

921
00:45:50,280 --> 00:45:53,640
and it is exactly the reason that the world is so complex,

922
00:45:53,640 --> 00:45:58,920
you cannot, there is no, who should be the arbiter of truth?

923
00:45:58,920 --> 00:46:01,000
It's sure as fuck is not going to be an LLM.

924
00:46:01,000 --> 00:46:01,560
Yeah.

925
00:46:01,560 --> 00:46:06,120
It's, and so you need, you need this open permissionless world

926
00:46:06,120 --> 00:46:07,800
that allows anyone to speak.

927
00:46:07,800 --> 00:46:09,880
Otherwise, including LLMs, including agents

928
00:46:09,880 --> 00:46:11,560
and including regular people and so on.

929
00:46:11,560 --> 00:46:13,640
Otherwise, we have no chance of figuring out what's true.

930
00:46:13,640 --> 00:46:14,680
Like, no chance at all.

931
00:46:14,680 --> 00:46:16,360
If there's only like two or three models

932
00:46:16,360 --> 00:46:17,800
and they determine what the truth is

933
00:46:17,800 --> 00:46:19,160
or like, there's one Skynet

934
00:46:19,160 --> 00:46:20,680
and it determines what the truth is.

935
00:46:20,680 --> 00:46:21,800
Like, we're in hell already.

936
00:46:21,800 --> 00:46:23,560
There's no, there's no difference.

937
00:46:23,560 --> 00:46:26,360
And, and I feel like that's what annoys me about the,

938
00:46:27,240 --> 00:46:29,560
all the AI bros and I listen to so many of them now.

939
00:46:29,560 --> 00:46:31,240
Like, I, I just want to puke.

940
00:46:31,240 --> 00:46:33,720
Like, my, my brain is basically throwing up constantly

941
00:46:33,720 --> 00:46:36,680
because I just, two X, all the AI bro talking

942
00:46:36,680 --> 00:46:39,160
over the last couple months into my brain.

943
00:46:39,160 --> 00:46:42,680
And they just, it's, it's, it's, it's very apparent to me

944
00:46:42,680 --> 00:46:46,520
that no one of them ever listened to any philosopher's ever.

945
00:46:46,520 --> 00:46:46,920
I don't know.

946
00:46:46,920 --> 00:46:49,800
And, and, and they actually figured some things out

947
00:46:49,800 --> 00:46:51,880
over the last couple thousand years, you know.

948
00:46:51,880 --> 00:46:54,280
So, and, and Silicon Valley is very dismissive

949
00:46:54,280 --> 00:46:54,920
of all that stuff.

950
00:46:54,920 --> 00:46:58,840
So, yeah, but, you know, talking about philosophy,

951
00:46:58,840 --> 00:47:01,880
I think that there's a very important topic to touch

952
00:47:01,880 --> 00:47:04,520
because we cannot forget

953
00:47:04,600 --> 00:47:07,320
fundamental questions in our modern times.

954
00:47:07,320 --> 00:47:10,360
Like, I'm talking about 50 years ago or something.

955
00:47:10,360 --> 00:47:13,640
We didn't have the penetration of the technology

956
00:47:13,640 --> 00:47:16,200
that we have to know this society.

957
00:47:16,200 --> 00:47:19,240
And we're still not answering some fundamental questions

958
00:47:19,240 --> 00:47:20,280
about that.

959
00:47:20,280 --> 00:47:23,240
And we kind of nightly use everything.

960
00:47:23,240 --> 00:47:27,080
Like, we, we kind of, we never understood

961
00:47:27,080 --> 00:47:29,560
what means to have some communication between people.

962
00:47:29,560 --> 00:47:30,280
Oh, yeah.

963
00:47:30,280 --> 00:47:32,120
I mean, we, we, we still don't know.

964
00:47:32,120 --> 00:47:33,320
And talking about something.

965
00:47:33,320 --> 00:47:34,360
Oh, what the internet is basically.

966
00:47:34,360 --> 00:47:34,760
Exactly.

967
00:47:34,760 --> 00:47:36,680
Like, and we are talking about things

968
00:47:36,680 --> 00:47:38,920
that now people take us totally normal.

969
00:47:38,920 --> 00:47:39,400
Yeah.

970
00:47:39,400 --> 00:47:42,520
And even, you know, like something required.

971
00:47:42,520 --> 00:47:44,760
I mean, you know, like, I mean, there, there is,

972
00:47:44,760 --> 00:47:47,400
it's very obvious that we are not well adapted to that.

973
00:47:47,400 --> 00:47:49,720
I mean, there's a reason why the suicide rate is like,

974
00:47:49,720 --> 00:47:52,040
as high as it ever was in the history of humanity,

975
00:47:52,040 --> 00:47:54,600
basically, that, that we have the meaning crisis.

976
00:47:54,600 --> 00:47:56,920
I mean, I mentioned Verveke all the time in, in conversation,

977
00:47:56,920 --> 00:48:00,760
but, but if you, if you ever want to learn about

978
00:48:00,760 --> 00:48:02,280
the meaning crisis, basically, like,

979
00:48:02,760 --> 00:48:03,640
that's your guy.

980
00:48:03,640 --> 00:48:07,960
He has like a, a 50 part lecture on why is it

981
00:48:07,960 --> 00:48:10,360
that everyone feels like hopeless and meaningless

982
00:48:10,360 --> 00:48:14,520
and all the people, you know, like, why do we have a math,

983
00:48:16,040 --> 00:48:17,960
problem worldwide, basically?

984
00:48:17,960 --> 00:48:21,800
Well, why do we have, yeah, a suicide epidemic

985
00:48:21,800 --> 00:48:22,840
and all those kinds of things.

986
00:48:22,840 --> 00:48:24,840
I think it relates to all this stuff where,

987
00:48:25,880 --> 00:48:27,720
and that's also, that's one thing that the eyebrows,

988
00:48:27,720 --> 00:48:28,600
for example, get wrong.

989
00:48:28,600 --> 00:48:31,480
I think it's, it's, we are,

990
00:48:32,040 --> 00:48:34,360
in some sense, disconnected from reality already.

991
00:48:35,000 --> 00:48:38,120
Um, just by, you know, being on the screens all the time,

992
00:48:38,120 --> 00:48:40,440
stroke, doom scrolling all the time and so on,

993
00:48:40,440 --> 00:48:43,160
not, not really touching grass, you know, like,

994
00:48:43,160 --> 00:48:43,880
we do it right now.

995
00:48:46,200 --> 00:48:49,560
Screen, literally touching grass right now.

996
00:48:49,560 --> 00:48:52,040
That's why this is outside, outside conversation.

997
00:48:52,040 --> 00:48:55,480
But, but, and, and again, like, this is not,

998
00:48:55,480 --> 00:48:57,400
these are not my arguments, it's all Verveke's arguments,

999
00:48:57,400 --> 00:48:59,800
but it's, it's, we, we lost touch with reality

1000
00:48:59,800 --> 00:49:01,640
and that is very, very unhealthy.

1001
00:49:01,640 --> 00:49:06,440
And the, the, the thing about LLMs is they were never

1002
00:49:06,440 --> 00:49:08,360
in touch with reality in the first place.

1003
00:49:08,360 --> 00:49:10,760
And so his whole argument is like,

1004
00:49:10,760 --> 00:49:13,480
to make them truly care about information,

1005
00:49:13,480 --> 00:49:14,520
we would have to embody them,

1006
00:49:14,520 --> 00:49:17,000
because why do you care about information?

1007
00:49:17,000 --> 00:49:19,320
It's basically, because I think I'm going to die.

1008
00:49:19,320 --> 00:49:21,240
Yeah, if you don't care about drinking or eating,

1009
00:49:21,240 --> 00:49:21,960
you're going to die.

1010
00:49:21,960 --> 00:49:24,120
So if you really, really thirsty, thirsty,

1011
00:49:24,120 --> 00:49:27,160
your whole salience landscape and your whole being

1012
00:49:27,160 --> 00:49:28,600
is about, I need to drink.

1013
00:49:29,080 --> 00:49:30,840
That's the, that's the information you're caring about.

1014
00:49:30,840 --> 00:49:32,040
And everything stems from there.

1015
00:49:32,040 --> 00:49:35,240
So basically, through human-like intelligence,

1016
00:49:35,240 --> 00:49:37,000
grounds out in being itself.

1017
00:49:37,000 --> 00:49:38,520
Yeah. And grounds out in embodiment

1018
00:49:38,520 --> 00:49:39,320
and all those kind of things.

1019
00:49:39,320 --> 00:49:41,560
Yeah. And the, the AI bros completely glides over that.

1020
00:49:41,560 --> 00:49:44,040
They're just like, yeah.

1021
00:49:44,040 --> 00:49:44,680
Well, you know.

1022
00:49:44,680 --> 00:49:46,840
I think that they are used to more people

1023
00:49:46,840 --> 00:49:48,840
that are assuming things as normal.

1024
00:49:48,840 --> 00:49:49,240
Yeah.

1025
00:49:49,240 --> 00:49:51,960
I got this, this, this, what they was talking about.

1026
00:49:51,960 --> 00:49:54,680
Also like, why, why, why do you want to spread out and conquer?

1027
00:49:54,680 --> 00:49:56,920
Like, first of all, not every human.

1028
00:49:56,920 --> 00:50:00,040
Like, a minority, a minority of humanity

1029
00:50:00,040 --> 00:50:01,720
wants to spread out and conquer everything.

1030
00:50:01,720 --> 00:50:02,040
Yeah.

1031
00:50:02,040 --> 00:50:04,440
A minority of like psychopaths, basically.

1032
00:50:04,440 --> 00:50:04,840
Yeah.

1033
00:50:04,840 --> 00:50:07,160
Most people are content, you know?

1034
00:50:07,160 --> 00:50:09,160
It's like, truly a minority of humans

1035
00:50:09,160 --> 00:50:11,400
wants to spread through the universe

1036
00:50:11,400 --> 00:50:12,920
and, you know, plant their seed everywhere

1037
00:50:12,920 --> 00:50:13,720
and conquer everything.

1038
00:50:13,720 --> 00:50:14,200
Yeah.

1039
00:50:14,200 --> 00:50:17,720
And all the AI people basically make the argument

1040
00:50:17,720 --> 00:50:19,320
that this is what the AI is going to do

1041
00:50:19,320 --> 00:50:21,240
because it wants to do, because it's the right thing to do.

1042
00:50:21,240 --> 00:50:26,280
And it's like, most, like, why is exploration,

1043
00:50:26,280 --> 00:50:28,440
why does it have such a high value in the first place?

1044
00:50:28,440 --> 00:50:29,960
Because scientists are explorers

1045
00:50:29,960 --> 00:50:32,760
and all the AI arguments are written by researchers

1046
00:50:32,760 --> 00:50:33,800
which are scientists, you know?

1047
00:50:33,800 --> 00:50:35,880
Like, why go out and find out new things?

1048
00:50:35,880 --> 00:50:36,520
What's the appeal?

1049
00:50:36,520 --> 00:50:37,960
Like, a tree doesn't do that.

1050
00:50:37,960 --> 00:50:38,440
Yeah.

1051
00:50:38,440 --> 00:50:41,080
But we would never argue that the tree is not a life, for example.

1052
00:50:41,080 --> 00:50:43,560
Or also, like a tree is very smart as well, you know?

1053
00:50:43,560 --> 00:50:44,200
It's like, of course.

1054
00:50:44,200 --> 00:50:48,680
And we wouldn't have a better smart shape for the environment.

1055
00:50:48,680 --> 00:50:51,480
Yeah, you're, you're adopted to the environment and so on.

1056
00:50:51,480 --> 00:50:54,360
And it's unclear to what kind of environment

1057
00:50:54,360 --> 00:50:57,160
the LLMs are adapted for because, again,

1058
00:50:57,160 --> 00:50:59,160
like, they're not really embodied in that sense.

1059
00:50:59,160 --> 00:51:03,000
And I mean, they're adapting right now for, you know, like,

1060
00:51:04,440 --> 00:51:08,760
if you want to be, I don't know, if you want to be generous,

1061
00:51:08,760 --> 00:51:11,320
then it's like a symbiotic relationship with humanity.

1062
00:51:12,120 --> 00:51:14,440
If you want to be, I don't know, more,

1063
00:51:15,960 --> 00:51:19,320
if you woke up with the doomer foot in you,

1064
00:51:19,320 --> 00:51:22,040
then it's like they're parasitic on humanity.

1065
00:51:22,040 --> 00:51:24,840
It's like, but yeah, I don't know.

1066
00:51:24,840 --> 00:51:26,760
Anyway, well, thanks for coming to my talk.

1067
00:51:27,800 --> 00:51:29,160
The coffee is clearly working.

1068
00:51:29,160 --> 00:51:29,400
Yeah.

1069
00:51:30,680 --> 00:51:31,640
It's very fun for me, too.

1070
00:51:32,360 --> 00:51:34,840
But I think that there is something very important to reflect

1071
00:51:36,120 --> 00:51:41,160
here because, like, the penetration of technology, as I was saying,

1072
00:51:41,160 --> 00:51:43,480
is something that is new, you know?

1073
00:51:43,480 --> 00:51:43,880
Yeah.

1074
00:51:43,880 --> 00:51:47,880
Like, we don't understand, we don't understand it yet as a society.

1075
00:51:47,880 --> 00:51:50,680
Yeah, experts that say, or, you know,

1076
00:51:52,040 --> 00:51:53,880
scalloped experts got it right.

1077
00:51:53,880 --> 00:51:54,840
Exactly.

1078
00:51:54,840 --> 00:51:55,560
I'm telling you.

1079
00:51:55,560 --> 00:51:56,440
Like, I was saying...

1080
00:51:56,440 --> 00:51:57,880
The older I get, the more,

1081
00:51:57,880 --> 00:51:59,880
also visually, the more I turn myself into that.

1082
00:51:59,880 --> 00:52:03,400
Yeah, they got the rights, definitely.

1083
00:52:04,440 --> 00:52:05,880
But I think that we are going to,

1084
00:52:06,760 --> 00:52:11,400
we are going to have to pass through through through all these answers

1085
00:52:12,280 --> 00:52:13,400
and questions.

1086
00:52:13,400 --> 00:52:15,400
Like, we are going to have to answer a lot of the questions

1087
00:52:15,400 --> 00:52:18,200
that we avoid to answer for more than 50 years.

1088
00:52:18,200 --> 00:52:19,720
Do you think we're going to make it?

1089
00:52:19,720 --> 00:52:21,080
I think that we are going to make it.

1090
00:52:21,080 --> 00:52:22,840
You're optimistic, in the end?

1091
00:52:22,840 --> 00:52:28,200
I'm bullish on philosophy, definitely.

1092
00:52:28,200 --> 00:52:29,560
I think that there is...

1093
00:52:29,560 --> 00:52:31,720
We are going to recover some fundamental questions

1094
00:52:31,720 --> 00:52:34,120
that we avoid to answer for so many years.

1095
00:52:34,920 --> 00:52:36,760
And that's because we are going to reach

1096
00:52:38,120 --> 00:52:42,840
some level of consciousness about the environment that we are living in

1097
00:52:42,840 --> 00:52:44,600
and the surroundings that we have around.

1098
00:52:45,720 --> 00:52:49,560
And, for example, I was saying yesterday, this...

1099
00:52:51,160 --> 00:52:53,960
...case of Olaskad, Sarah with...

1100
00:52:54,840 --> 00:52:55,240
I don't know.

1101
00:52:55,240 --> 00:52:56,920
This woman that is against Facebook,

1102
00:52:58,120 --> 00:53:01,800
Meta, that he sees revealing all these details.

1103
00:53:01,800 --> 00:53:03,240
Didn't she also write...

1104
00:53:03,240 --> 00:53:04,440
Didn't she also...

1105
00:53:05,800 --> 00:53:08,760
I think she sued Pornhub with the same number?

1106
00:53:08,760 --> 00:53:10,120
Well, I don't know, baby.

1107
00:53:12,040 --> 00:53:13,480
I used to write the bell, but I'm not sure.

1108
00:53:14,040 --> 00:53:16,040
But did you say Sarah White or something like that?

1109
00:53:16,040 --> 00:53:19,320
Sarah White, Sarah White something more.

1110
00:53:20,040 --> 00:53:21,240
I don't remember right now.

1111
00:53:21,240 --> 00:53:22,440
I'm very bad for names.

1112
00:53:22,440 --> 00:53:23,560
But I'll try to find...

1113
00:53:23,560 --> 00:53:27,400
Because there is a great book on it on the Pornhub thing,

1114
00:53:27,400 --> 00:53:28,760
but it might be something else entirely.

1115
00:53:28,760 --> 00:53:30,680
It was published three days ago.

1116
00:53:30,680 --> 00:53:31,320
That was a...

1117
00:53:32,680 --> 00:53:39,400
But she, in this declaration that she do,

1118
00:53:39,960 --> 00:53:43,240
she talks about whole Facebook or Meta now,

1119
00:53:43,960 --> 00:53:51,560
they kind of leverage their ability to spy into the world,

1120
00:53:51,560 --> 00:53:53,320
sorry, into the life of people.

1121
00:53:53,320 --> 00:53:54,840
Because, for example, they treat...

1122
00:53:54,840 --> 00:53:57,000
Like, if you are a teenager,

1123
00:53:57,000 --> 00:54:02,280
a teenager of 15 years old or 15 between 13 and 17,

1124
00:54:02,280 --> 00:54:04,200
and you, for example, upload a selfie,

1125
00:54:04,760 --> 00:54:06,600
and then you delete the selfie,

1126
00:54:06,600 --> 00:54:08,040
they take this as a signal...

1127
00:54:08,040 --> 00:54:08,600
Oh, of course.

1128
00:54:08,600 --> 00:54:10,040
...of the kind of depression...

1129
00:54:10,040 --> 00:54:10,760
Yeah, sure.

1130
00:54:10,760 --> 00:54:11,640
...or bad mood.

1131
00:54:11,640 --> 00:54:14,120
Targeted advertising for the mood boosters.

1132
00:54:14,120 --> 00:54:15,080
Yeah, let's go.

1133
00:54:15,080 --> 00:54:19,160
So, which means that the advertising industry

1134
00:54:19,160 --> 00:54:21,880
is going to try to make you down all the time.

1135
00:54:21,880 --> 00:54:22,440
Sure, sure.

1136
00:54:22,440 --> 00:54:23,240
So, how do you think?

1137
00:54:23,240 --> 00:54:23,640
Sure.

1138
00:54:23,640 --> 00:54:25,160
Because it's more efficient.

1139
00:54:25,160 --> 00:54:28,200
It's more effective for them to sell you things with your depressed.

1140
00:54:28,200 --> 00:54:31,320
I don't even think that this was the plan from the get-go.

1141
00:54:31,320 --> 00:54:34,200
I don't think all these companies were founded too.

1142
00:54:34,200 --> 00:54:38,120
Hey, let's do this and let's make all the 14-year-old teenage girls

1143
00:54:38,120 --> 00:54:40,920
depressed and then sell them pills and whatever.

1144
00:54:41,000 --> 00:54:42,520
I don't think there was the plan.

1145
00:54:42,520 --> 00:54:48,760
But once you build the system and you do this data-driven advertisement campaigns, basically,

1146
00:54:48,760 --> 00:54:50,680
this is what works best, you know?

1147
00:54:50,680 --> 00:54:54,840
And it's like, you don't even stand back and reflect and ask like,

1148
00:54:54,840 --> 00:54:55,640
why are we doing this?

1149
00:54:55,640 --> 00:54:58,600
It's just like, okay, we have to hit the quarterly profit margins

1150
00:54:58,600 --> 00:55:02,600
and we have to, you know, we are obliged to maximize shareholder value

1151
00:55:02,600 --> 00:55:03,800
and all this kind of stuff.

1152
00:55:03,800 --> 00:55:06,520
And then you just do all these things and it's like, okay,

1153
00:55:06,520 --> 00:55:08,600
if we tweak this little knob here,

1154
00:55:08,680 --> 00:55:13,080
then we'll increase our revenue by like, you know, 10% or whatever.

1155
00:55:13,080 --> 00:55:14,040
And that's how it happens.

1156
00:55:14,040 --> 00:55:15,880
It's just like, it's hell man.

1157
00:55:15,880 --> 00:55:22,040
Like when I was listening to this woman, like the first thing much that appeared in my mind

1158
00:55:22,040 --> 00:55:27,000
was, you know, Harry Potter, the mentors, like these kind of spirits

1159
00:55:27,000 --> 00:55:30,200
that suit your bad energy and they can sit by that.

1160
00:55:30,200 --> 00:55:30,920
Yeah, absolutely.

1161
00:55:30,920 --> 00:55:33,400
So that was the first thing much that appeared in my mind,

1162
00:55:33,400 --> 00:55:35,320
just when I was hitting this woman.

1163
00:55:35,320 --> 00:55:35,800
Yeah.

1164
00:55:35,880 --> 00:55:40,360
Yeah, as a fact, like a whole whole effect, we as a society,

1165
00:55:40,360 --> 00:55:49,640
we let entities like these take us as I manage our emotions, our state.

1166
00:55:49,640 --> 00:55:50,600
And that's the thing.

1167
00:55:50,600 --> 00:55:53,400
That's that's also what most people kind of gloss over,

1168
00:55:53,400 --> 00:55:56,040
like bringing it back to the LLMs and all those kind of things.

1169
00:55:56,040 --> 00:55:59,240
It's a corporation like Facebook or like Google or what have you,

1170
00:55:59,240 --> 00:56:02,520
is already alien intelligence.

1171
00:56:02,520 --> 00:56:02,840
Yeah.

1172
00:56:02,840 --> 00:56:04,440
It's a non-human intelligence.

1173
00:56:04,440 --> 00:56:05,000
Yeah.

1174
00:56:05,000 --> 00:56:08,600
Like a city or a corporation or a state or what have you.

1175
00:56:09,240 --> 00:56:12,360
It's already a superhuman intelligence.

1176
00:56:12,360 --> 00:56:13,560
It just is.

1177
00:56:13,560 --> 00:56:17,720
And so we can deal with those as well in some sense, you know?

1178
00:56:17,720 --> 00:56:21,960
Like they are not necessarily destroying us all and so on.

1179
00:56:21,960 --> 00:56:27,160
I mean, in the Facebook and Google a little bit, but you get my point.

1180
00:56:27,160 --> 00:56:31,560
It's like, people forgot that we already have these agentic systems

1181
00:56:31,560 --> 00:56:33,640
that are bigger than a single human.

1182
00:56:33,720 --> 00:56:35,720
And we learn to kind of deal with them.

1183
00:56:35,720 --> 00:56:36,760
I think we suck at it.

1184
00:56:36,760 --> 00:56:38,280
I think we're really bad at it right now.

1185
00:56:38,280 --> 00:56:40,120
I think we used to be better in the past.

1186
00:56:40,120 --> 00:56:40,360
Yeah.

1187
00:56:40,360 --> 00:56:43,080
Because in the past, everything we had is spirit.

1188
00:56:43,080 --> 00:56:53,480
We knew how to like, we were able to deal with this better because we knew that these things existed.

1189
00:56:53,480 --> 00:56:53,880
Yeah.

1190
00:56:53,880 --> 00:56:59,000
There is such a thing like, I don't know, the spirit of anger or the spirit of lust.

1191
00:56:59,000 --> 00:57:00,520
And you know how to deal with it, for example.

1192
00:57:00,520 --> 00:57:03,160
Like it's its own entity that does things.

1193
00:57:03,160 --> 00:57:06,680
And if you're possessed by a spirit, then that's how you act.

1194
00:57:06,680 --> 00:57:07,160
And so on.

1195
00:57:07,160 --> 00:57:09,560
And it's like, I think we see this with the Facebook engineers, for example.

1196
00:57:09,560 --> 00:57:13,000
It's like, if you're possessed by Facebook and you're sitting in there

1197
00:57:13,000 --> 00:57:17,240
and you use your computer science degree to optimize the click-through rate of like,

1198
00:57:17,240 --> 00:57:19,880
you know, like depression pose for 40 year olds,

1199
00:57:19,880 --> 00:57:21,960
like you were possessed by the spirit of Facebook.

1200
00:57:21,960 --> 00:57:22,840
And that's not a good thing.

1201
00:57:23,560 --> 00:57:25,080
And yeah, well, anyway.

1202
00:57:26,040 --> 00:57:28,040
So DVM's are going to fix them.

1203
00:57:28,920 --> 00:57:32,760
Not that I'm making a freedom technology, I'm going to fix that.

1204
00:57:34,280 --> 00:57:37,720
But it is actually like another book recommendation since you mentioned Harry Potter.

1205
00:57:38,920 --> 00:57:41,560
Most people don't know about the book, but it is a great Bitcoin book.

1206
00:57:42,280 --> 00:57:45,320
It's Momo by Michel Enne.

1207
00:57:45,320 --> 00:57:47,320
It's a German book, a German author.

1208
00:57:47,320 --> 00:57:48,360
But it was translated.

1209
00:57:48,840 --> 00:57:54,040
And it's basically, there are similar entities like the Harry Potter spirits that you mentioned

1210
00:57:54,040 --> 00:57:59,160
that suck that's the mentors that suck stuff out of you and so on.

1211
00:57:59,560 --> 00:58:06,760
And for this book, for Momo, it's the gray man, like they look like gray man in suits,

1212
00:58:06,760 --> 00:58:07,880
and they're time thieves.

1213
00:58:08,760 --> 00:58:11,000
So for me, conceptually, it's inflation.

1214
00:58:13,000 --> 00:58:16,920
It's basically the bankers that are inflating the currency and they are stealing your time.

1215
00:58:17,640 --> 00:58:20,760
That's what the book basically, those are the villains of the book.

1216
00:58:20,760 --> 00:58:25,320
And if you read the book through a Bitcoin lens, it's fantastic, it's like it.

1217
00:58:25,320 --> 00:58:26,520
Everything makes perfect sense.

1218
00:58:27,560 --> 00:58:28,680
Yeah, so a great book.

1219
00:58:29,640 --> 00:58:38,200
I, something that you say, like we was more kind of conscious some time ago that we are now.

1220
00:58:38,840 --> 00:58:41,800
Sapping is a more conscious action than like watching the TikTok scene.

1221
00:58:41,800 --> 00:58:42,120
Yeah, yeah.

1222
00:58:42,120 --> 00:58:46,680
No, but you're talking about the technology and the penetration of it in our lives.

1223
00:58:46,680 --> 00:58:53,320
Yeah, I remember when I was a kid, an internet arrived to my home,

1224
00:58:54,440 --> 00:58:59,640
and my parents was so worried, because I should not upload nothing.

1225
00:58:59,640 --> 00:59:02,440
Did the internet still make sounds or you're quite a bit,

1226
00:59:03,320 --> 00:59:06,680
I think you're quite a bit younger than me, but it was still like you could listen to the internet.

1227
00:59:06,680 --> 00:59:10,680
Yeah, I could listen to the internet through the phone.

1228
00:59:10,680 --> 00:59:11,160
Fantastic.

1229
00:59:11,960 --> 00:59:16,680
But my parents were so worried, because you know, you should be careful to

1230
00:59:16,680 --> 00:59:20,760
not upload nothing, any personal information, photographs.

1231
00:59:21,400 --> 00:59:26,760
And now it's like a son normalize that everyone is putting photos of their assets.

1232
00:59:26,760 --> 00:59:29,080
They were ahead of the curve.

1233
00:59:29,080 --> 00:59:32,760
Is that why you are privacy conscious because of your parents?

1234
00:59:32,760 --> 00:59:39,320
They normally, I don't know, they are probably, but now they are normalizing it.

1235
00:59:39,400 --> 00:59:40,200
Oh, yeah, sure.

1236
00:59:40,200 --> 00:59:41,640
You know, it's as everyone.

1237
00:59:41,640 --> 00:59:42,680
It's completely normalized.

1238
00:59:42,680 --> 00:59:43,800
You're a weird one.

1239
00:59:43,800 --> 00:59:45,560
I'm always the weird one when I say like,

1240
00:59:46,120 --> 00:59:48,760
you know, please don't take pictures of my face, for example.

1241
00:59:52,600 --> 00:59:55,640
It was a lot of work to not have my face on the internet.

1242
00:59:55,640 --> 00:59:57,400
So don't ruin it in an instant, please.

1243
00:59:59,400 --> 01:00:04,040
Yeah, I mean, that is a fucking big discussion that we're going to have to do also.

1244
01:00:04,040 --> 01:00:07,480
Yeah, all the privacy implications of this new technology.

1245
01:00:07,960 --> 01:00:11,000
I don't think we're going to make it in that sense.

1246
01:00:11,000 --> 01:00:12,200
I think it's just going to be normal.

1247
01:00:12,200 --> 01:00:16,360
And now with the generative models, for example, it's just like, I mean,

1248
01:00:18,120 --> 01:00:22,440
revenge porn has a new dimension now with a generative porn.

1249
01:00:23,240 --> 01:00:26,920
But it's not all bad, because you have plausible deniability now,

1250
01:00:26,920 --> 01:00:30,440
where you can say there's actually a real photo.

1251
01:00:30,440 --> 01:00:33,800
It was like, no, no, it's clearly, clearly generated.

1252
01:00:33,800 --> 01:00:34,680
It wasn't me.

1253
01:00:34,680 --> 01:00:35,240
It wasn't me.

1254
01:00:35,240 --> 01:00:36,920
I would never do anything with a donkey.

1255
01:00:37,800 --> 01:00:39,160
I didn't grow that coat.

1256
01:00:39,160 --> 01:00:39,880
What's an LMM?

1257
01:00:40,760 --> 01:00:41,320
Exactly.

1258
01:00:42,440 --> 01:00:42,920
Exactly.

1259
01:00:42,920 --> 01:00:44,760
That's a great excuse for every single body.

1260
01:00:44,760 --> 01:00:45,080
Yeah.

1261
01:00:45,080 --> 01:00:46,760
It's like, oh, that was vive-coded.

1262
01:00:46,760 --> 01:00:48,200
That was vive-coded.

1263
01:00:48,200 --> 01:00:49,080
I promise.

1264
01:00:49,080 --> 01:00:52,040
I wasn't even conscious about what's Dell M was right in there.

1265
01:00:52,840 --> 01:00:54,520
Just put Dell M in Dell.

1266
01:00:57,800 --> 01:00:58,840
Yeah, we'll see how it goes.

1267
01:00:58,840 --> 01:01:05,160
I mean, I'm always optimistic that in the grand scheme of things, humanity will figure it out.

1268
01:01:05,160 --> 01:01:09,960
Like, we will figure it out how to deal with all these things.

1269
01:01:09,960 --> 01:01:11,960
But it is already very confusing.

1270
01:01:11,960 --> 01:01:14,360
And it's going to get more confusing going forward.

1271
01:01:14,360 --> 01:01:17,640
I mean, I still have some normal friends and like, my family as well.

1272
01:01:18,600 --> 01:01:21,000
Quite like, my mom is very normal, for example.

1273
01:01:21,000 --> 01:01:23,400
It's like, she sends me,

1274
01:01:24,440 --> 01:01:26,840
they're always the same fake stuff all the time.

1275
01:01:26,840 --> 01:01:29,000
Not realizing it's all fake, basically.

1276
01:01:29,000 --> 01:01:31,640
And it's just like, oh, man, bless your soul.

1277
01:01:32,200 --> 01:01:35,640
It's crazy, it's crazy.

1278
01:01:35,640 --> 01:01:37,320
So this is what just gets worse and worse.

1279
01:01:37,320 --> 01:01:38,840
Yeah, it's going to get to go to...

1280
01:01:38,840 --> 01:01:42,680
And I think that, as I say, like, we're going to have to answer a lot of questions.

1281
01:01:42,680 --> 01:01:43,480
Oh, yeah, absolutely.

1282
01:01:43,480 --> 01:01:46,520
Because it's going to be a nightmare, you know?

1283
01:01:46,520 --> 01:01:49,240
Like, all these new wearables.

1284
01:01:50,120 --> 01:01:52,920
Like, Android DX is releasing.

1285
01:01:52,920 --> 01:01:55,400
It's a partnership between Google and Samsung.

1286
01:01:55,400 --> 01:01:59,320
They're releasing the new Android for a mental reality.

1287
01:02:00,120 --> 01:02:03,080
And everyone is going to be wearing one of these glasses.

1288
01:02:03,080 --> 01:02:05,000
And you're not going to be able to say nothing to them.

1289
01:02:05,000 --> 01:02:07,160
Like, they are going to use be recording every time.

1290
01:02:07,160 --> 01:02:08,040
Let's see, let's see.

1291
01:02:08,040 --> 01:02:09,160
I'm not sure.

1292
01:02:09,160 --> 01:02:11,960
Like, the last time this was tried with Google Glass.

1293
01:02:12,600 --> 01:02:14,520
And everyone caught them, the glass holes.

1294
01:02:14,520 --> 01:02:15,800
And everyone hated it.

1295
01:02:15,800 --> 01:02:17,880
And so it didn't really pan out the last time.

1296
01:02:17,880 --> 01:02:20,840
But now, you know, like, now it looks like the regular sunglasses.

1297
01:02:20,840 --> 01:02:23,800
So probably there is no turning back.

1298
01:02:23,800 --> 01:02:26,120
But yeah, we'll see.

1299
01:02:26,120 --> 01:02:29,880
I think one thing that you mentioned in terms of philosophical questions,

1300
01:02:29,880 --> 01:02:35,000
kind of what is real, I think that's one of the most interesting ones,

1301
01:02:35,000 --> 01:02:36,120
because it's so hard to answer.

1302
01:02:37,560 --> 01:02:40,840
And but also people really crave realness.

1303
01:02:41,560 --> 01:02:48,200
And I think in the monetary sense, Bitcoin is the most real money period.

1304
01:02:48,200 --> 01:02:51,080
There is so like, it's even more real and gold in some sense.

1305
01:02:51,080 --> 01:02:55,880
And, you know, everyone that hasn't understood Bitcoin deeply will completely frown at

1306
01:02:55,960 --> 01:02:59,880
the statement and like, you know, like basically, oh, what are the crazy people talking about?

1307
01:02:59,880 --> 01:03:00,920
But it is actually true.

1308
01:03:00,920 --> 01:03:01,720
Like, it's the most

1309
01:03:04,120 --> 01:03:06,120
realness is like, what is very viable in some sense?

1310
01:03:06,600 --> 01:03:08,840
And it is way more very viable than gold.

1311
01:03:08,840 --> 01:03:10,200
Because you don't know how much gold there is.

1312
01:03:10,200 --> 01:03:12,120
And like, you have to start mining all the other things.

1313
01:03:12,120 --> 01:03:17,320
Maybe we, you know, drill down the earth so much more and we'll just like quadruple our

1314
01:03:17,320 --> 01:03:18,920
gold supply in an instant, what have you.

1315
01:03:18,920 --> 01:03:21,080
You can't do all that with Bitcoin.

1316
01:03:21,080 --> 01:03:23,880
And Bitcoin also, it bottoms out in physical reality.

1317
01:03:23,880 --> 01:03:25,320
That's what's so interesting about it.

1318
01:03:25,320 --> 01:03:32,120
It's actually, you can literally know like what had to happen for this Bitcoin to exist and so on.

1319
01:03:32,120 --> 01:03:34,600
And so it's all traceable and verifiable and so on.

1320
01:03:34,600 --> 01:03:35,960
That's what makes Bitcoin so special.

1321
01:03:35,960 --> 01:03:39,800
So in the monetary sense, Bitcoin is the realest money we ever had.

1322
01:03:39,800 --> 01:03:44,120
And people really care about reality and people really also care about money.

1323
01:03:44,120 --> 01:03:51,480
And the other side of reality is like, like, what is authentic, you know?

1324
01:03:51,480 --> 01:03:53,320
Like the opposite of fake, basically.

1325
01:03:54,280 --> 01:04:01,240
But not necessarily in the verifiability sense, but in the saying what you truly mean,

1326
01:04:01,240 --> 01:04:03,640
sense, in some sense, you know, like a big question.

1327
01:04:03,640 --> 01:04:08,360
Yeah, no, but it's because there's all these aspects, realness and what is real.

1328
01:04:08,360 --> 01:04:10,440
Again, people really care about realness.

1329
01:04:10,440 --> 01:04:11,880
They really, really do.

1330
01:04:11,880 --> 01:04:21,320
And if you are in a kind of many, many environment, like if you're on a platform

1331
01:04:21,320 --> 01:04:24,440
and you're at the edge of being the platform, always,

1332
01:04:24,440 --> 01:04:27,160
you don't even have a way to express yourself fully.

1333
01:04:27,160 --> 01:04:28,120
You just don't.

1334
01:04:28,120 --> 01:04:31,880
So you're just going to lie for the reason of not being the platform.

1335
01:04:31,880 --> 01:04:33,720
Or lie is a strong word.

1336
01:04:33,720 --> 01:04:34,840
But also that.

1337
01:04:35,640 --> 01:04:37,320
I mean, the COVID times show this clearly.

1338
01:04:38,680 --> 01:04:40,200
You're not going to say what you think.

1339
01:04:40,760 --> 01:04:42,200
You're not going to say what you think.

1340
01:04:42,200 --> 01:04:43,800
That's it, that's what I'm going to say.

1341
01:04:43,800 --> 01:04:47,240
And so once we are in a situation where people are not saying what they think,

1342
01:04:47,240 --> 01:04:50,520
how can we have any chance of finding out what's real and what's true?

1343
01:04:51,080 --> 01:04:55,800
And so that's why I care so much about Noster, the free expression side.

1344
01:04:55,800 --> 01:04:57,800
Because I do really care about reality.

1345
01:04:57,800 --> 01:05:01,560
And I also, even more than that in some sense,

1346
01:05:01,560 --> 01:05:07,400
I care about our pathway and our tools of finding out what really is actually.

1347
01:05:08,280 --> 01:05:12,840
What is the best approach to get a grip and a handle on reality?

1348
01:05:12,840 --> 01:05:16,760
And so looping it back to closed platforms, closed systems,

1349
01:05:16,760 --> 01:05:22,360
and closed LLMs, and all those kind of things, we won't be able to figure out

1350
01:05:22,360 --> 01:05:25,160
what's real in a closed paradigm.

1351
01:05:25,160 --> 01:05:25,800
We just won't.

1352
01:05:26,600 --> 01:05:28,200
Because of all the things I just mentioned.

1353
01:05:31,400 --> 01:05:34,600
You'll have to lie, you'll have to, you'll have all these chilling effects,

1354
01:05:34,600 --> 01:05:35,320
all those kind of things.

1355
01:05:35,320 --> 01:05:36,680
You're going to be constrained, you're going to be afraid.

1356
01:05:36,680 --> 01:05:38,040
Exactly, you have all these constraints.

1357
01:05:38,040 --> 01:05:38,680
Yeah, yeah.

1358
01:05:38,680 --> 01:05:40,280
Comvisionate, yeah, yeah.

1359
01:05:41,800 --> 01:05:44,920
And also long term, like the other thing is also, if you

1360
01:05:45,800 --> 01:05:51,560
basically, if you need to box it in, the energy required to do the policing,

1361
01:05:51,560 --> 01:05:53,320
the range energy from the system.

1362
01:05:53,320 --> 01:05:54,440
So long term, it's not even stable.

1363
01:05:54,440 --> 01:05:58,760
That's why all authoritarian regimes basically fail, give them a long enough time frame.

1364
01:05:58,760 --> 01:06:01,880
But anyway, thanks for coming to my second talk.

1365
01:06:01,880 --> 01:06:03,400
That was a very strong coffee, man.

1366
01:06:05,160 --> 01:06:07,240
For some many facts, it was a good cycle idea.

1367
01:06:08,680 --> 01:06:08,840
Yeah.

1368
01:06:08,840 --> 01:06:12,520
So what's your plans the next couple of weeks and months for DBMCP?

1369
01:06:13,160 --> 01:06:17,640
So I'm going to keep exploring the spec.

1370
01:06:17,640 --> 01:06:26,040
I was having, as we are here in Madrid and some, I was just kissing with Hazard.

1371
01:06:26,040 --> 01:06:27,400
Thanks for geo-doxing me.

1372
01:06:27,400 --> 01:06:28,440
No, no, it's okay.

1373
01:06:28,440 --> 01:06:29,400
Everyone knows, everyone knows.

1374
01:06:29,400 --> 01:06:31,480
I think that you mentioned this a couple of times.

1375
01:06:31,480 --> 01:06:32,040
Everyone knows.

1376
01:06:33,880 --> 01:06:40,040
But yeah, I think I'm going to keep exploring the specification,

1377
01:06:40,120 --> 01:06:49,240
and so on, because DBN spec is something with the design that works to be received.

1378
01:06:49,240 --> 01:06:52,040
Yeah, it wasn't really designed with all the alarms in mind, right?

1379
01:06:52,040 --> 01:06:55,240
Or with this stuff in mind.

1380
01:06:56,600 --> 01:06:59,480
What would you revise?

1381
01:07:00,360 --> 01:07:03,240
Ah, so because the spec is very generic, right?

1382
01:07:03,240 --> 01:07:07,480
Yeah, yeah, not just that, it's like very bloated.

1383
01:07:07,480 --> 01:07:12,200
Because every, like, the current state of DBMs on the spec

1384
01:07:13,000 --> 01:07:17,800
says that you have a job request per kind.

1385
01:07:17,800 --> 01:07:20,440
So just segregate the job request per kind.

1386
01:07:20,440 --> 01:07:22,520
So you have a kind for text summarization.

1387
01:07:22,520 --> 01:07:23,560
Yeah, yeah, I see.

1388
01:07:23,560 --> 01:07:25,000
You have a kind for imagination.

1389
01:07:25,000 --> 01:07:27,480
All your transcripts, what have you, and all those kinds of things.

1390
01:07:27,480 --> 01:07:31,640
And all of them have a very different schemas for the inputs,

1391
01:07:31,640 --> 01:07:33,320
and the output that you spread from that.

1392
01:07:33,880 --> 01:07:37,000
So it really makes the spec really bloated.

1393
01:07:37,000 --> 01:07:41,400
So if you want to support everything, you're going to have a great time

1394
01:07:43,000 --> 01:07:44,200
trying to implement all of that.

1395
01:07:45,400 --> 01:07:49,560
And I think that there is a limitation.

1396
01:07:51,400 --> 01:07:53,400
So how would you revise it?

1397
01:07:54,040 --> 01:07:55,720
Don't listen to the semantic web people.

1398
01:07:55,720 --> 01:07:56,440
That's my own input.

1399
01:07:57,640 --> 01:08:00,600
I was having a great discussion with Hasa about this,

1400
01:08:00,600 --> 01:08:05,000
and I think that he wrote this article yesterday, I think.

1401
01:08:05,960 --> 01:08:09,080
About how DBMs are broken, basically.

1402
01:08:09,640 --> 01:08:12,200
This came from some discussions that we have.

1403
01:08:14,360 --> 01:08:20,200
But yeah, payments for DBMs are also a big topic because

1404
01:08:21,640 --> 01:08:25,560
now you have a lot of different possibilities to make payments

1405
01:08:25,560 --> 01:08:27,720
like with the Nossa where I connect.

1406
01:08:30,120 --> 01:08:32,680
It's not clear how you should pay a DBM.

1407
01:08:33,160 --> 01:08:37,480
Some of the spec that you set up, the job request,

1408
01:08:37,480 --> 01:08:41,960
or some event in the middle of the process of interacting with the DBM,

1409
01:08:42,520 --> 01:08:48,360
other spec to use a Nossa, and it's not really clear.

1410
01:08:48,360 --> 01:08:51,080
And the specification that tried to clarify that,

1411
01:08:51,720 --> 01:08:54,280
I was talking also a lot with Don't Believe the hype.

1412
01:08:55,720 --> 01:08:59,880
And he's the most active guy working on the DBMs.

1413
01:08:59,880 --> 01:09:00,760
He's the DBM guy.

1414
01:09:00,840 --> 01:09:07,560
Also dashing and very good people, but he was trying to clarify this,

1415
01:09:07,560 --> 01:09:16,200
and he wrote an extension for the NIP to clarify the new parameters that DBMs have,

1416
01:09:16,200 --> 01:09:17,320
or he's working with them.

1417
01:09:18,280 --> 01:09:22,040
I never got much, so it's pretty unclear how you should proceed

1418
01:09:22,040 --> 01:09:24,280
if you want to create and work with DBMs.

1419
01:09:25,480 --> 01:09:28,760
So I think that works to revisit all of this.

1420
01:09:31,080 --> 01:09:38,600
Also, I think that we should use more the ephemeral kinds for DBMs,

1421
01:09:38,600 --> 01:09:44,440
because right now a sign running DBNCP doesn't have to stick around forever.

1422
01:09:44,440 --> 01:09:44,920
Exactly.

1423
01:09:45,720 --> 01:09:48,040
I'm running for DBNCP and running the website,

1424
01:09:48,040 --> 01:09:53,080
and I'm also running a relay for the backend, let's say.

1425
01:09:55,480 --> 01:09:57,560
I don't care in the majority of events.

1426
01:09:58,120 --> 01:10:01,160
Once they accomplish their mission.

1427
01:10:01,160 --> 01:10:05,640
I mean, most of the things are very simple kind of job requests,

1428
01:10:05,640 --> 01:10:09,160
and what's the job is done, you don't care about the input event anymore,

1429
01:10:09,160 --> 01:10:10,280
or what have you, those kind of things.

1430
01:10:11,800 --> 01:10:18,120
Also, something that I really like of exploring with my implementation of DBNCP

1431
01:10:18,120 --> 01:10:23,560
is that I induce using a single kind for everything.

1432
01:10:24,200 --> 01:10:26,120
Instead of, because it makes sense.

1433
01:10:26,120 --> 01:10:32,040
I mean, because it's all the natural language prompts anyway, right, in some sense.

1434
01:10:32,040 --> 01:10:35,560
Yeah, no, it's because DB, sorry, MCP,

1435
01:10:36,280 --> 01:10:39,720
have the protocol is specified.

1436
01:10:40,440 --> 01:10:41,320
Do you have a method?

1437
01:10:41,320 --> 01:10:41,720
Yeah, okay.

1438
01:10:41,720 --> 01:10:44,040
Everything is an MCP tool, so it's all the same kind.

1439
01:10:44,040 --> 01:10:44,760
Yeah, yeah, yeah.

1440
01:10:44,760 --> 01:10:48,120
And they are using a pattern that is carried reflection.

1441
01:10:48,920 --> 01:10:52,200
So they have this method that is called list tools.

1442
01:10:52,600 --> 01:10:53,800
Watch out, we're getting rolled over.

1443
01:10:55,880 --> 01:10:58,040
So they have this method that is carried...

1444
01:10:58,040 --> 01:11:00,280
You would think you're safe in the park, but no?

1445
01:11:00,280 --> 01:11:02,680
No, no, no, it's like, it's basically...

1446
01:11:02,680 --> 01:11:05,480
You can't get hit by a track in any moment.

1447
01:11:08,520 --> 01:11:13,000
But yeah, I mean, MCP have this reflection pattern,

1448
01:11:13,000 --> 01:11:16,200
which means that you have a method that is carried list tools,

1449
01:11:16,200 --> 01:11:20,520
and when you do list tools, you are going to get all the schema

1450
01:11:21,320 --> 01:11:24,600
that you need to build your request.

1451
01:11:25,320 --> 01:11:29,320
So you don't need a previously schema,

1452
01:11:29,320 --> 01:11:32,600
or previously shared schema or something like that.

1453
01:11:32,600 --> 01:11:37,240
You have a reflection, which means that the schema is part of the response.

1454
01:11:37,240 --> 01:11:37,720
Exactly.

1455
01:11:37,720 --> 01:11:39,000
Like you get...

1456
01:11:39,000 --> 01:11:40,840
So you don't need to redefine the schema.

1457
01:11:40,840 --> 01:11:41,400
Exactly.

1458
01:11:41,960 --> 01:11:45,800
You get the schema of the inputs, and the output is also standardized.

1459
01:11:45,880 --> 01:11:49,800
So you get what you need to make the request,

1460
01:11:49,800 --> 01:11:51,960
and to understand what you expect from it.

1461
01:11:52,680 --> 01:11:56,200
So that makes that every tool can fit in the same can,

1462
01:11:56,200 --> 01:12:01,240
because beforehand you can understand how to input the data,

1463
01:12:01,240 --> 01:12:02,680
or how to build the request.

1464
01:12:02,680 --> 01:12:04,040
Yeah, I see, I see.

1465
01:12:04,040 --> 01:12:05,640
Yeah, that makes a ton of sense.

1466
01:12:09,480 --> 01:12:13,400
Both things are very new, but TVMs predate MCP tools.

1467
01:12:14,360 --> 01:12:18,680
And so we didn't know of a way to have like a general purpose.

1468
01:12:19,320 --> 01:12:25,080
DVM that does all the things using LLMs, basically using MCP.

1469
01:12:25,080 --> 01:12:28,360
So now we have that that makes a lot of sense.

1470
01:12:28,360 --> 01:12:33,320
Yeah, but I also think that there are a space for predifying kin.

1471
01:12:33,320 --> 01:12:34,200
Oh, absolutely.

1472
01:12:34,200 --> 01:12:35,320
I mean, not all...

1473
01:12:36,760 --> 01:12:38,360
Absolutely, because you...

1474
01:12:39,560 --> 01:12:42,360
I mean, if you want to add a watermark to an image or whatever,

1475
01:12:42,760 --> 01:12:44,040
you're not going to use an LLM.

1476
01:12:44,040 --> 01:12:45,880
You're going to use image magic.

1477
01:12:45,880 --> 01:12:46,440
Yeah.

1478
01:12:46,440 --> 01:12:51,320
Or like for some trends, you know, like transcoding or whatever,

1479
01:12:51,320 --> 01:12:53,320
like converting one file type to another one,

1480
01:12:53,320 --> 01:12:55,080
you're going to use FFmpeg or what have you.

1481
01:12:55,080 --> 01:12:56,760
Yeah, those kind of things.

1482
01:12:56,760 --> 01:12:59,560
It's, I mean, again, like with MCP tools,

1483
01:12:59,560 --> 01:13:03,560
it's still, there's this layer in between where you have the natural language,

1484
01:13:05,080 --> 01:13:06,840
adapter, so to speak, for these tools.

1485
01:13:06,840 --> 01:13:08,280
But it's not always needed.

1486
01:13:08,280 --> 01:13:11,480
Like, I mean, I think a lot of DVM,

1487
01:13:11,480 --> 01:13:15,400
I mean, the main idea of DVM supposed to actually be like a vending machine.

1488
01:13:15,400 --> 01:13:17,400
Like if it's something that's very, very stupid.

1489
01:13:17,400 --> 01:13:19,720
Take a coin, give me something.

1490
01:13:22,200 --> 01:13:26,440
And I think we, you know, regular super-stupid vending machines

1491
01:13:26,440 --> 01:13:28,440
are still extremely useful.

1492
01:13:28,440 --> 01:13:28,840
Yeah.

1493
01:13:28,840 --> 01:13:31,000
And yeah, I think we'll just have a pause.

1494
01:13:31,000 --> 01:13:31,880
Yeah, definitely.

1495
01:13:31,880 --> 01:13:35,640
But something to reflect about this is that, you know,

1496
01:13:35,640 --> 01:13:41,160
it's virtually impossible that you can come up with an integer for every job request,

1497
01:13:41,240 --> 01:13:45,880
because the job request can be so customized and so personalized.

1498
01:13:46,520 --> 01:13:49,800
Like how the fact you define, I don't know,

1499
01:13:49,800 --> 01:13:52,440
a kind for a specific job that I have.

1500
01:13:52,440 --> 01:13:57,400
I mean, we had this problem already with ports, you know, like with the port system,

1501
01:13:57,960 --> 01:14:00,040
where you have some numbers that are reserved.

1502
01:14:00,040 --> 01:14:01,400
But if you build a new application,

1503
01:14:01,400 --> 01:14:04,040
you need to come up with a new port number if it's network.

1504
01:14:04,040 --> 01:14:04,520
Yeah.

1505
01:14:04,520 --> 01:14:07,960
And then you're just like, you know, there is many integers,

1506
01:14:07,960 --> 01:14:10,120
but that doesn't make it less messy.

1507
01:14:10,120 --> 01:14:13,160
It's like they are, they are infinite integers.

1508
01:14:13,160 --> 01:14:17,320
But they are also an infinite amount of possibilities.

1509
01:14:17,320 --> 01:14:19,240
So there is more infinite real numbers.

1510
01:14:19,240 --> 01:14:20,040
Did you know that?

1511
01:14:20,040 --> 01:14:20,360
Oh, yeah.

1512
01:14:21,480 --> 01:14:22,840
Because they can't be negative.

1513
01:14:25,320 --> 01:14:26,360
Fucking set theory.

1514
01:14:28,360 --> 01:14:28,920
Great.

1515
01:14:28,920 --> 01:14:30,440
Oh, mathematicians are so smart.

1516
01:14:30,440 --> 01:14:32,680
Oh, there's like 10 different kinds of infinity.

1517
01:14:32,680 --> 01:14:34,680
The one infinity is bigger than the other infinity.

1518
01:14:36,520 --> 01:14:38,200
Oh, well, anyway, it's a ball.

1519
01:14:38,920 --> 01:14:41,400
Yeah, no, I think cleaning that up would make a lot of sense.

1520
01:14:41,400 --> 01:14:44,840
And then I think it's just about building the tools

1521
01:14:44,840 --> 01:14:49,560
and make it really easy for other people to create the MCP tools, for example,

1522
01:14:49,560 --> 01:14:53,160
or also making it really easy to create custom DBMs.

1523
01:14:53,160 --> 01:14:53,400
Yeah.

1524
01:14:54,280 --> 01:14:57,880
No, I'm really glad also for the work that I'm doing in the MCP,

1525
01:14:57,880 --> 01:15:05,400
because I think that is the most, the easiest way to create a DBM out of a MCP tool,

1526
01:15:05,400 --> 01:15:07,000
that MCP is its own protocol.

1527
01:15:07,480 --> 01:15:10,920
So if you have a MCP server, you have a VBM.

1528
01:15:11,480 --> 01:15:12,520
Yeah, yeah, yeah.

1529
01:15:12,520 --> 01:15:14,360
And you don't need to write any line of code.

1530
01:15:14,360 --> 01:15:15,720
Yeah, yeah, basically.

1531
01:15:15,720 --> 01:15:16,280
Yeah.

1532
01:15:16,280 --> 01:15:19,480
And that makes the library that I'm writing, writing,

1533
01:15:19,480 --> 01:15:21,720
and the packages that I'm writing, so lightweight.

1534
01:15:22,440 --> 01:15:27,800
And so like a focus on the specific task of the communication,

1535
01:15:27,800 --> 01:15:33,000
and the city handling, handling different stuff in the DBM thing.

1536
01:15:33,080 --> 01:15:36,440
And I don't have to bloat my packages,

1537
01:15:36,440 --> 01:15:38,280
because I have to support this on the other thing.

1538
01:15:38,280 --> 01:15:38,600
Yeah.

1539
01:15:38,600 --> 01:15:43,480
I just support the DBMCP concept is that you have a client,

1540
01:15:43,480 --> 01:15:46,440
an MCP client that connects to the MCP server.

1541
01:15:47,080 --> 01:15:50,920
And you kind of grab that into a DBM,

1542
01:15:50,920 --> 01:15:53,080
which means that you are going to be listening to a relay

1543
01:15:53,880 --> 01:15:57,800
and reacting to some events that can come from that relay.

1544
01:15:57,800 --> 01:15:59,640
And you also can publish things on the relay.

1545
01:16:00,280 --> 01:16:02,920
So it's like a DBM wrapper for a DBMCP.

1546
01:16:03,800 --> 01:16:07,320
That make it so specific, you know,

1547
01:16:07,320 --> 01:16:10,840
so that I don't have to support any or the kinds of

1548
01:16:11,560 --> 01:16:13,560
every use case or one note.

1549
01:16:13,560 --> 01:16:14,840
Like at the MCP server,

1550
01:16:14,840 --> 01:16:18,280
have its own computation and do its own thing.

1551
01:16:19,160 --> 01:16:23,640
And the package that I'm building just do the rest.

1552
01:16:23,640 --> 01:16:27,560
That is the networking part or the handling part of the,

1553
01:16:27,640 --> 01:16:28,440
with no set.

1554
01:16:28,440 --> 01:16:29,080
Yeah.

1555
01:16:29,080 --> 01:16:30,120
No, awesome, awesome.

1556
01:16:30,120 --> 01:16:32,600
I think the most powerful thing,

1557
01:16:32,600 --> 01:16:34,600
what I saw in the last couple of weeks is,

1558
01:16:34,600 --> 01:16:37,000
again, I think it has to be really easy for someone to

1559
01:16:38,440 --> 01:16:41,000
publish their things to no stir.

1560
01:16:41,000 --> 01:16:43,640
So it can appear in the marketplaces

1561
01:16:43,640 --> 01:16:45,480
and the things we talked about.

1562
01:16:45,480 --> 01:16:47,160
And what makes it so powerful is that

1563
01:16:47,720 --> 01:16:50,200
like people can just pick it up and use it immediately

1564
01:16:50,200 --> 01:16:51,000
if they want to.

1565
01:16:51,000 --> 01:16:51,160
Yeah.

1566
01:16:51,160 --> 01:16:52,920
And we saw this in some of the demos,

1567
01:16:52,920 --> 01:16:54,680
where just, you know,

1568
01:16:54,680 --> 01:16:56,360
some of the, again, coming back to some of the things

1569
01:16:56,360 --> 01:16:58,200
that Pablo has been working on where,

1570
01:16:59,000 --> 01:17:01,800
you know, to try to vibe code and app, basically.

1571
01:17:01,800 --> 01:17:03,880
And it fails because the agent is confused.

1572
01:17:03,880 --> 01:17:09,640
And someone else publishes a code snippet or a specific MCP tool.

1573
01:17:10,280 --> 01:17:11,960
And without doing anything,

1574
01:17:11,960 --> 01:17:14,600
the agent kind of tries again and it fixes itself.

1575
01:17:14,600 --> 01:17:18,360
Because now it has the more power,

1576
01:17:18,360 --> 01:17:19,080
the more knowledge.

1577
01:17:20,040 --> 01:17:20,360
Nice.

1578
01:17:20,360 --> 01:17:22,200
I think that is so cool.

1579
01:17:22,200 --> 01:17:24,040
I think it's so powerful because also

1580
01:17:24,680 --> 01:17:28,920
kind of allow everyone to expose their computational resources

1581
01:17:28,920 --> 01:17:31,080
or assets like you have,

1582
01:17:32,040 --> 01:17:33,880
we can, as we were talking before,

1583
01:17:33,880 --> 01:17:36,920
we can take the IP address that you have as an asset, you know.

1584
01:17:36,920 --> 01:17:37,400
Yeah.

1585
01:17:37,400 --> 01:17:37,880
Yeah.

1586
01:17:37,880 --> 01:17:38,360
Yeah.

1587
01:17:38,920 --> 01:17:41,720
You are a person that uses BPN, BPS.

1588
01:17:41,720 --> 01:17:42,680
Yeah, BPN, in fact.

1589
01:17:43,880 --> 01:17:47,160
You don't, you cannot watch YouTube anymore

1590
01:17:47,720 --> 01:17:49,000
in a good way.

1591
01:17:49,000 --> 01:17:53,800
But maybe I'm a person that I have a IP address

1592
01:17:54,120 --> 01:17:55,480
from ISP.

1593
01:17:56,120 --> 01:17:57,400
And I don't care about BPN,

1594
01:17:57,400 --> 01:17:58,840
so I can connect to YouTube,

1595
01:17:58,840 --> 01:18:00,200
it's a one-out-rate limit.

1596
01:18:00,200 --> 01:18:03,640
So I can create a DBNCP or a NCP tool

1597
01:18:03,640 --> 01:18:05,160
that dole lots things from YouTube

1598
01:18:05,160 --> 01:18:06,840
and put it on blossom, for example.

1599
01:18:07,560 --> 01:18:10,520
And I can do that with any restriction

1600
01:18:10,520 --> 01:18:12,440
and sell that as a kind of assets.

1601
01:18:12,440 --> 01:18:12,920
Yeah.

1602
01:18:12,920 --> 01:18:15,240
We saw with one of the experiments that

1603
01:18:16,920 --> 01:18:18,840
B actually built it from Ziggyt,

1604
01:18:18,840 --> 01:18:21,320
where you can do it for

1605
01:18:22,200 --> 01:18:24,120
like any kind of HTTP request,

1606
01:18:24,120 --> 01:18:25,320
which was super interesting.

1607
01:18:25,320 --> 01:18:25,560
Yeah.

1608
01:18:25,560 --> 01:18:26,760
Where it's just, you know,

1609
01:18:26,760 --> 01:18:29,560
like I'm willing to execute

1610
01:18:30,280 --> 01:18:32,520
the your HTTP request for you

1611
01:18:32,520 --> 01:18:33,880
using my IP, basically.

1612
01:18:34,440 --> 01:18:38,120
And I'll give you back the result and so on.

1613
01:18:38,120 --> 01:18:40,120
And I mean, I think we'll see all of that.

1614
01:18:40,120 --> 01:18:42,760
If this type of experimentation continues,

1615
01:18:42,760 --> 01:18:46,120
we saw also the atomic swaps, for example,

1616
01:18:46,120 --> 01:18:47,640
with the snorvers, the answers and so on.

1617
01:18:47,640 --> 01:18:48,200
That's great.

1618
01:18:48,200 --> 01:18:49,000
Yeah, great name.

1619
01:18:49,640 --> 01:18:50,440
Ask.

1620
01:18:52,120 --> 01:18:53,000
Ask protocols.

1621
01:18:54,200 --> 01:18:58,760
And you could use that to, again,

1622
01:18:58,760 --> 01:19:01,160
like execute all of that atomically.

1623
01:19:01,160 --> 01:19:01,400
Yeah.

1624
01:19:01,400 --> 01:19:05,080
It's basically, at the same time,

1625
01:19:05,080 --> 01:19:08,040
you get back the result and receive the payment.

1626
01:19:08,040 --> 01:19:08,600
Yeah.

1627
01:19:08,600 --> 01:19:10,920
And like I only pay you if I actually

1628
01:19:12,840 --> 01:19:14,680
can make sure that I get the correct result

1629
01:19:14,680 --> 01:19:15,800
and those kinds of things.

1630
01:19:15,800 --> 01:19:18,440
So yeah, I mean, it opens up a whole new area

1631
01:19:18,840 --> 01:19:19,480
of possibilities.

1632
01:19:19,480 --> 01:19:19,800
Yeah.

1633
01:19:19,800 --> 01:19:21,400
It's very, it's quite simple.

1634
01:19:21,400 --> 01:19:23,240
Building blocks, basically, in some sense.

1635
01:19:23,240 --> 01:19:24,360
That's a beautiful part, yeah.

1636
01:19:25,320 --> 01:19:26,680
I mean, well, that's simple.

1637
01:19:26,680 --> 01:19:28,360
It's at least it's very, it's,

1638
01:19:28,360 --> 01:19:29,720
I think it's very understandable.

1639
01:19:29,720 --> 01:19:29,880
Yeah.

1640
01:19:29,880 --> 01:19:31,080
It's like once you, once you grab.

1641
01:19:31,080 --> 01:19:31,800
A broad table.

1642
01:19:31,800 --> 01:19:32,120
Yeah.

1643
01:19:33,000 --> 01:19:35,000
There's no hidden source in, like,

1644
01:19:35,000 --> 01:19:36,200
snoring signatures or whatever.

1645
01:19:36,200 --> 01:19:38,920
Like you can actually look at it and understand it and so on.

1646
01:19:38,920 --> 01:19:40,280
And again, like coming back to the

1647
01:19:41,000 --> 01:19:42,760
permissionless nature of it all.

1648
01:19:42,760 --> 01:19:45,640
You can just use it and run with it and

1649
01:19:45,640 --> 01:19:46,600
build stuff on top of it.

1650
01:19:46,600 --> 01:19:48,200
So I know it's bound to happen.

1651
01:19:48,600 --> 01:19:51,080
And I think that these are the building blocks

1652
01:19:51,080 --> 01:19:55,720
that really make us closer to this level of collaboration

1653
01:19:55,720 --> 01:19:57,240
that really make us free.

1654
01:19:57,240 --> 01:20:00,760
Because if we can collaborate at these levels.

1655
01:20:00,760 --> 01:20:01,400
Yeah.

1656
01:20:01,400 --> 01:20:03,720
More market-based, less silo-based.

1657
01:20:03,720 --> 01:20:05,240
There is nothing that can stop us.

1658
01:20:05,880 --> 01:20:07,640
I can, nothing can stop me to,

1659
01:20:08,920 --> 01:20:12,200
to sell my IP for, to someone if it needs.

1660
01:20:12,200 --> 01:20:15,720
And I can also benefit from that.

1661
01:20:16,200 --> 01:20:18,440
Very simple things like, again,

1662
01:20:18,440 --> 01:20:21,240
selling your idle CPU or GPU power.

1663
01:20:21,240 --> 01:20:21,560
Yeah.

1664
01:20:21,560 --> 01:20:23,800
If someone wants to, you know, train models

1665
01:20:23,800 --> 01:20:25,240
to do whatever and those kind of things.

1666
01:20:25,240 --> 01:20:25,400
Yeah.

1667
01:20:25,400 --> 01:20:27,400
Like, I mean, those are all very old ideas.

1668
01:20:27,400 --> 01:20:29,160
Like, city at home, those kind of things.

1669
01:20:29,160 --> 01:20:31,080
They don't have that protein folding,

1670
01:20:31,080 --> 01:20:32,920
like, had a big boost with that and so on.

1671
01:20:32,920 --> 01:20:33,480
Yeah.

1672
01:20:33,480 --> 01:20:35,240
But it's all, it's all like,

1673
01:20:35,240 --> 01:20:38,200
you actually as an, it's not really a marketplace.

1674
01:20:38,200 --> 01:20:40,520
It's like very, very big companies,

1675
01:20:40,520 --> 01:20:45,000
very big institutions doing that and trying to get everyone aboard.

1676
01:20:45,000 --> 01:20:47,880
But I think once you flip it to a marketplace kind of scenario,

1677
01:20:49,080 --> 01:20:51,800
the very interesting things are going to happen.

1678
01:20:51,800 --> 01:20:53,720
I think we'll have a lot of innovation,

1679
01:20:53,720 --> 01:20:55,000
a lot of good stuff come from that again.

1680
01:20:55,000 --> 01:20:57,640
Like, coming back to the previous example of like,

1681
01:20:57,640 --> 01:20:58,840
browser extensions and so on.

1682
01:20:58,840 --> 01:21:00,040
It's like, basically, yeah.

1683
01:21:00,040 --> 01:21:01,560
All my, the browsers that I use,

1684
01:21:01,560 --> 01:21:04,200
they're just, they're kind of like a compilation

1685
01:21:04,200 --> 01:21:05,320
of extensions in some sense.

1686
01:21:05,320 --> 01:21:08,680
It's like, if I can't install my like,

1687
01:21:08,680 --> 01:21:10,360
six, seven favorite extensions,

1688
01:21:10,360 --> 01:21:13,640
I don't even want to use the browser because it's useless to me.

1689
01:21:13,880 --> 01:21:15,400
Yeah, yeah, yeah.

1690
01:21:15,400 --> 01:21:18,040
I mean, I'm, I'm at least going to need an ostracina, you know,

1691
01:21:18,040 --> 01:21:19,320
in my browser.

1692
01:21:19,320 --> 01:21:21,800
If I can't install that, I can't use the browser for anything.

1693
01:21:21,800 --> 01:21:22,360
Yeah.

1694
01:21:22,360 --> 01:21:22,920
Yeah.

1695
01:21:22,920 --> 01:21:24,040
That's the thing that it's worth.

1696
01:21:24,040 --> 01:21:28,200
And I think that this is because we now kind of have the right building blocks,

1697
01:21:28,200 --> 01:21:28,840
as you were saying.

1698
01:21:29,800 --> 01:21:32,440
And these ideas are not new, as you were saying.

1699
01:21:33,080 --> 01:21:35,080
But the building blocks that we have now,

1700
01:21:35,080 --> 01:21:36,440
yes, they are new.

1701
01:21:37,000 --> 01:21:38,920
And they have a lot of potential.

1702
01:21:38,920 --> 01:21:39,720
Definitely.

1703
01:21:39,720 --> 01:21:41,800
And we have existing infrastructure to plug in.

1704
01:21:41,800 --> 01:21:42,680
Yeah, we have Bitcoin.

1705
01:21:42,680 --> 01:21:43,240
We have no start.

1706
01:21:43,240 --> 01:21:44,200
We have those things.

1707
01:21:44,200 --> 01:21:44,680
Exactly.

1708
01:21:44,680 --> 01:21:47,000
And you can just, should we cross over to the other park?

1709
01:21:47,000 --> 01:21:47,480
Let's go.

1710
01:21:47,480 --> 01:21:52,920
Do one more, one more circle and, and, and with maybe not another coffee,

1711
01:21:52,920 --> 01:21:54,680
or there will be a third cutout,

1712
01:21:54,680 --> 01:21:56,680
but then a nice orange juice or something.

1713
01:21:56,680 --> 01:21:57,800
Yeah, I don't just, it's nice.

1714
01:21:58,760 --> 01:21:59,320
Ah, wow.

1715
01:21:59,320 --> 01:22:01,240
This bag is very beautiful to you.

1716
01:22:01,240 --> 01:22:03,960
Oh, yeah, I just, no, it's all very pretty.

1717
01:22:03,960 --> 01:22:06,840
And now, now it's officially spring, I think, more or less.

1718
01:22:06,840 --> 01:22:07,880
This weather is beautiful.

1719
01:22:07,880 --> 01:22:08,600
Yeah, it's great.

1720
01:22:08,600 --> 01:22:10,680
We're not like shit weather in comparison.

1721
01:22:10,760 --> 01:22:11,720
It's comfortable.

1722
01:22:12,360 --> 01:22:14,840
You stop the angles when the weather gets better.

1723
01:22:16,120 --> 01:22:17,240
It was because of all the nerds.

1724
01:22:18,200 --> 01:22:19,880
The nerds brought the bad weather.

1725
01:22:22,120 --> 01:22:24,200
What do you think of the last six weeks?

1726
01:22:24,200 --> 01:22:25,560
What did you get out of it?

1727
01:22:25,560 --> 01:22:26,520
I enjoyed so much.

1728
01:22:27,240 --> 01:22:28,760
I mean, you're a veteran already.

1729
01:22:30,280 --> 01:22:33,400
For me, it was very, it was the most enjoyable cohort,

1730
01:22:33,400 --> 01:22:34,760
I think, this time around, because,

1731
01:22:36,200 --> 01:22:37,480
no, but I did the least.

1732
01:22:37,480 --> 01:22:38,040
Yeah.

1733
01:22:38,040 --> 01:22:38,680
Yeah.

1734
01:22:40,360 --> 01:22:41,800
I did nothing but vibe code.

1735
01:22:41,800 --> 01:22:42,360
That's great.

1736
01:22:43,080 --> 01:22:45,240
But I think that was super, super intense.

1737
01:22:46,120 --> 01:22:48,440
And part of the intensity was,

1738
01:22:48,440 --> 01:22:50,200
because now we have these new tools

1739
01:22:50,200 --> 01:22:54,120
that makes everything faster and drier and better cheaper.

1740
01:22:54,120 --> 01:22:54,440
Yeah.

1741
01:22:54,440 --> 01:22:56,120
So you can do way more,

1742
01:22:56,120 --> 01:22:57,480
with way less, in some sense.

1743
01:22:57,480 --> 01:23:00,520
Which means that even if the effort that you have to put

1744
01:23:00,520 --> 01:23:05,080
into make the things is less, the mental effort is still

1745
01:23:05,080 --> 01:23:07,320
the same level or more,

1746
01:23:07,320 --> 01:23:08,840
because you have to think faster,

1747
01:23:09,400 --> 01:23:11,880
to resolve the issues faster and so on.

1748
01:23:12,440 --> 01:23:14,920
So that makes everything super intense.

1749
01:23:14,920 --> 01:23:15,160
Yeah.

1750
01:23:15,160 --> 01:23:16,840
You don't have one project anymore.

1751
01:23:16,840 --> 01:23:19,080
You have like 15 vibe code and projects.

1752
01:23:20,040 --> 01:23:23,080
And that is another question to answer.

1753
01:23:23,080 --> 01:23:23,160
Yeah.

1754
01:23:23,160 --> 01:23:24,120
But that's the thing.

1755
01:23:24,120 --> 01:23:27,640
Coming back to your point about the philosophy

1756
01:23:27,640 --> 01:23:29,080
becoming more important.

1757
01:23:29,080 --> 01:23:31,800
And we touched on this in past conversation as well.

1758
01:23:31,800 --> 01:23:33,640
But it's like, if you can't build anything,

1759
01:23:33,640 --> 01:23:36,440
basically, the question of what to build

1760
01:23:36,440 --> 01:23:38,360
and why it becomes way more important.

1761
01:23:38,360 --> 01:23:39,000
Definitely.

1762
01:23:39,000 --> 01:23:39,240
Yeah.

1763
01:23:39,240 --> 01:23:40,600
It's like what you can build.

1764
01:23:40,600 --> 01:23:41,000
Absolutely.

1765
01:23:41,000 --> 01:23:43,320
And let's say this stuff works perfectly

1766
01:23:43,320 --> 01:23:46,440
and you can vibe code anything into existence,

1767
01:23:46,440 --> 01:23:48,600
basically almost in an instant for free.

1768
01:23:48,600 --> 01:23:49,320
What do you do?

1769
01:23:49,320 --> 01:23:52,760
Like, why are you actually building the thing that you?

1770
01:23:52,760 --> 01:23:53,800
That's a question.

1771
01:23:53,800 --> 01:23:55,480
That's the right question, I think.

1772
01:23:55,480 --> 01:24:00,520
Because if, yeah, when you have a magic machine

1773
01:24:00,520 --> 01:24:01,560
that do everything for you,

1774
01:24:03,400 --> 01:24:04,680
now you have to choose what to do.

1775
01:24:05,320 --> 01:24:06,440
I'm going to do a good one.

1776
01:24:06,440 --> 01:24:07,240
Hey, guys.

1777
01:24:07,240 --> 01:24:08,840
And the Austrians, by the way,

1778
01:24:08,840 --> 01:24:10,280
figured this out, the Austrian economists,

1779
01:24:10,280 --> 01:24:13,000
is like, I don't remember the author anymore.

1780
01:24:13,000 --> 01:24:14,920
But time, the ultimate resource,

1781
01:24:14,920 --> 01:24:16,360
because they figured out that like,

1782
01:24:16,360 --> 01:24:19,160
even if such a thing would exist hypothetically,

1783
01:24:19,160 --> 01:24:20,840
you still have time constraints.

1784
01:24:20,840 --> 01:24:21,320
Yeah.

1785
01:24:21,320 --> 01:24:23,480
You still have to spend your time and attention on the things.

1786
01:24:23,480 --> 01:24:25,400
And so that's basically, you know,

1787
01:24:25,400 --> 01:24:27,400
that was the rebuttal to the Star Trek universe.

1788
01:24:27,400 --> 01:24:29,960
What if we have free energy and replicators?

1789
01:24:29,960 --> 01:24:31,800
Everyone has free energy.

1790
01:24:31,800 --> 01:24:33,160
Everyone has a replicator.

1791
01:24:33,160 --> 01:24:35,240
You still have to answer the question,

1792
01:24:35,240 --> 01:24:36,200
what to replicate.

1793
01:24:36,200 --> 01:24:36,360
Yeah.

1794
01:24:36,360 --> 01:24:37,640
And that's a philosophical question.

1795
01:24:37,640 --> 01:24:37,880
Yeah.

1796
01:24:37,880 --> 01:24:39,800
And it relates to values,

1797
01:24:39,800 --> 01:24:40,920
and what kind of values you have,

1798
01:24:40,920 --> 01:24:42,360
and what you actually care about.

1799
01:24:42,360 --> 01:24:42,520
Yeah.

1800
01:24:42,520 --> 01:24:43,960
And you can't automate that away.

1801
01:24:43,960 --> 01:24:46,120
So that's kind of, that's, yeah.

1802
01:24:46,120 --> 01:24:49,880
That's what we will have to focus on now in some sense.

1803
01:24:49,880 --> 01:24:51,560
I mean, you always have to focus on that,

1804
01:24:51,560 --> 01:24:55,160
but now it's even more pronounced, I feel like.

1805
01:24:55,160 --> 01:24:55,640
Yeah.

1806
01:24:55,640 --> 01:24:58,280
You know, the perfect sentence for this is,

1807
01:24:59,080 --> 01:25:01,320
a great power comes with a great responsibility.

1808
01:25:01,320 --> 01:25:01,640
Yeah.

1809
01:25:01,640 --> 01:25:02,600
Exactly.

1810
01:25:02,600 --> 01:25:03,640
It's quite a man.

1811
01:25:03,640 --> 01:25:04,280
Right.

1812
01:25:04,280 --> 01:25:05,080
The Uncle Ben.

1813
01:25:06,040 --> 01:25:07,000
The legendary one.

1814
01:25:07,560 --> 01:25:08,120
Uncle Ben.

1815
01:25:08,120 --> 01:25:11,000
Yeah, it's still totally true.

1816
01:25:11,000 --> 01:25:12,440
Like, oh, absolutely.

1817
01:25:12,440 --> 01:25:14,120
Like, and now you have these

1818
01:25:14,120 --> 01:25:16,920
so powerful tools that they're crazy, man.

1819
01:25:16,920 --> 01:25:19,720
Like, we were talking about the instant communication,

1820
01:25:19,720 --> 01:25:22,360
that is something that I answered.

1821
01:25:22,360 --> 01:25:24,360
But now we have this technology that is

1822
01:25:24,360 --> 01:25:25,960
fucking crazy, what you can do.

1823
01:25:25,960 --> 01:25:28,040
Instant Russian development team in your pocket.

1824
01:25:28,040 --> 01:25:28,440
Yeah.

1825
01:25:28,440 --> 01:25:28,840
Yeah.

1826
01:25:28,840 --> 01:25:29,400
Yeah.

1827
01:25:29,400 --> 01:25:32,200
That's, it's fucking in the imaginary world.

1828
01:25:33,080 --> 01:25:34,600
But now what?

1829
01:25:34,600 --> 01:25:35,160
Now what?

1830
01:25:36,280 --> 01:25:38,360
Now you have to answer this question of what?

1831
01:25:38,360 --> 01:25:40,040
What are you going to vibe code today?

1832
01:25:40,040 --> 01:25:40,520
Yeah.

1833
01:25:40,520 --> 01:25:41,080
Right.

1834
01:25:41,080 --> 01:25:41,560
Right.

1835
01:25:41,560 --> 01:25:42,120
Right.

1836
01:25:42,120 --> 01:25:43,320
That's right.

1837
01:25:43,320 --> 01:25:43,800
And wing.

1838
01:25:44,760 --> 01:25:45,080
All right.

1839
01:25:45,080 --> 01:25:46,360
Let's maybe stay at this side.

1840
01:25:46,360 --> 01:25:48,840
There's like a chainsaw on the other side.

1841
01:25:48,840 --> 01:25:49,960
So let's not.

1842
01:25:50,920 --> 01:25:51,320
All right.

1843
01:25:51,320 --> 01:25:51,720
Let's not.

1844
01:25:51,720 --> 01:25:52,120
Let's not.

1845
01:25:55,480 --> 01:25:57,320
Let's maybe stay away from the chainsaw.

1846
01:25:58,920 --> 01:25:59,880
The rest of the whole.

1847
01:26:00,520 --> 01:26:02,440
I was like, oh, let's walk the parks.

1848
01:26:02,440 --> 01:26:04,520
It's going to be nice and quiet and so on.

1849
01:26:04,520 --> 01:26:06,440
The one park you got rolled over by a truck

1850
01:26:06,440 --> 01:26:07,960
gave the second park there's a chainsaw.

1851
01:26:09,080 --> 01:26:10,200
Oh, well.

1852
01:26:10,200 --> 01:26:11,000
City sounds.

1853
01:26:11,000 --> 01:26:11,960
That's how it goes.

1854
01:26:11,960 --> 01:26:12,680
That's how it goes.

1855
01:26:13,560 --> 01:26:17,720
But I think that we touched some interesting points

1856
01:26:17,720 --> 01:26:19,400
and very important points.

1857
01:26:19,960 --> 01:26:23,000
Because we are going to have to answer all of this

1858
01:26:24,040 --> 01:26:25,080
in the next years.

1859
01:26:25,080 --> 01:26:25,720
Oh, absolutely.

1860
01:26:25,720 --> 01:26:27,480
And start to think about that.

1861
01:26:27,960 --> 01:26:30,040
To make us more prepared for all of that.

1862
01:26:30,760 --> 01:26:33,000
Because it's not something like a weekend avoid.

1863
01:26:33,000 --> 01:26:33,960
We can't avoid it.

1864
01:26:35,000 --> 01:26:36,600
No, you can't run away from this.

1865
01:26:36,600 --> 01:26:39,800
No, it's going to it's going to touch your life in one way or another.

1866
01:26:40,360 --> 01:26:44,520
And you're going to have to put yourself into the light and answer that.

1867
01:26:44,520 --> 01:26:46,280
Even though we like ostriches,

1868
01:26:46,920 --> 01:26:52,120
putting your head into the sand is not going to be the best strategy going forward.

1869
01:26:52,120 --> 01:26:55,400
That can work for 50 years or something like that, but not forever.

1870
01:26:56,360 --> 01:27:01,400
And now we are reaching the point of that is critical, you know.

1871
01:27:02,120 --> 01:27:04,360
And I think that there is, we,

1872
01:27:05,320 --> 01:27:07,880
like right now the penetration of this technology is not

1873
01:27:08,920 --> 01:27:12,760
the total penetration that we are going to see of this technology.

1874
01:27:12,760 --> 01:27:17,640
And like big companies are going to be about the neck.

1875
01:27:18,280 --> 01:27:21,480
And what I mean with this is that when everyone,

1876
01:27:22,360 --> 01:27:24,760
like this technology gets really democratized.

1877
01:27:24,760 --> 01:27:28,120
And every company and every business or as little as they are.

1878
01:27:29,560 --> 01:27:32,360
They are going to need this technology to to operate.

1879
01:27:35,800 --> 01:27:38,520
You know, they're not going to be a node offered,

1880
01:27:38,520 --> 01:27:40,920
enough offer to to supply the demand.

1881
01:27:41,880 --> 01:27:45,640
Because because these big providers are going to be a bottleneck.

1882
01:27:45,640 --> 01:27:48,680
I mean, now everything is like, you know, heavily VC funded.

1883
01:27:48,680 --> 01:27:50,120
Oh yeah, VC subsidized.

1884
01:27:50,120 --> 01:27:54,120
You see this clearly if you like the difference between the difference in cost

1885
01:27:54,120 --> 01:27:59,160
and use the cost between using cursor and using root code is massive.

1886
01:28:01,560 --> 01:28:03,080
So yeah, no, I agree.

1887
01:28:03,080 --> 01:28:07,400
I mean, I think it will also get cheaper with technological progress and so on.

1888
01:28:09,480 --> 01:28:13,560
But I think that the technical part is not just the only bottleneck.

1889
01:28:14,120 --> 01:28:14,920
Oh, absolutely.

1890
01:28:14,920 --> 01:28:18,520
The regulatory parts is also a big bottleneck.

1891
01:28:18,520 --> 01:28:22,920
Like a, what define an Iranian guy that runs to access.

1892
01:28:22,920 --> 01:28:25,240
We saw it already with Google, right?

1893
01:28:25,240 --> 01:28:26,680
Like it's a bottleneck for search.

1894
01:28:26,680 --> 01:28:30,840
Like again, Google defines what the truth is from the point of view of Google.

1895
01:28:30,840 --> 01:28:32,040
What are you allowed to find?

1896
01:28:32,040 --> 01:28:33,320
What are you allowed to search for?

1897
01:28:33,320 --> 01:28:34,520
All those kinds of things.

1898
01:28:34,520 --> 01:28:36,600
And if the political climate gets weird enough,

1899
01:28:36,600 --> 01:28:40,120
then it's just like stuff will just fucking not show up and disappear.

1900
01:28:40,120 --> 01:28:41,160
And all those kinds of things.

1901
01:28:41,240 --> 01:28:43,080
And it was a very simple thing in terms of,

1902
01:28:44,040 --> 01:28:46,120
basically, I mean, Google is not a simple thing anymore.

1903
01:28:46,120 --> 01:28:50,360
But when compared to LLMs, basically like plain text search in some sense.

1904
01:28:50,360 --> 01:28:50,760
Yeah.

1905
01:28:50,760 --> 01:28:55,160
And LLMs, it's like, you know, what are the LLMs allowed to think in some sense?

1906
01:28:55,160 --> 01:28:55,320
Yeah.

1907
01:28:55,320 --> 01:28:56,920
That's going to be the question.

1908
01:28:56,920 --> 01:28:57,080
Yeah.

1909
01:28:57,080 --> 01:28:57,800
So I agree.

1910
01:28:57,800 --> 01:29:03,480
It is like every central four spy definition is a bottleneck.

1911
01:29:03,480 --> 01:29:03,960
Yeah.

1912
01:29:03,960 --> 01:29:04,360
Yeah, yeah.

1913
01:29:04,360 --> 01:29:07,160
Every central thing is a bottleneck by definition.

1914
01:29:07,160 --> 01:29:08,120
So that's true.

1915
01:29:08,120 --> 01:29:08,520
That's true.

1916
01:29:09,240 --> 01:29:09,880
That's true.

1917
01:29:09,880 --> 01:29:15,160
And I learned anything that the role of us, like people that develops over open standards

1918
01:29:15,160 --> 01:29:19,160
and try to understand and realize what is this new technology

1919
01:29:19,160 --> 01:29:24,600
and how that can be applied to our lives and how they can be deployed

1920
01:29:24,600 --> 01:29:27,720
in a way that makes sense, it's so important.

1921
01:29:27,720 --> 01:29:29,880
And it's going to be so important in the future.

1922
01:29:29,880 --> 01:29:36,680
Because when you have the 80% of businesses wanting to use this technology

1923
01:29:36,760 --> 01:29:40,200
to operate and to create their business plans

1924
01:29:40,200 --> 01:29:42,920
or their marketing plans or campaigns or anything,

1925
01:29:42,920 --> 01:29:48,680
or they want to outsource part of the workload into this business thing,

1926
01:29:49,560 --> 01:29:54,440
okay, it's going to be a lot of a big bottleneck because this, you know,

1927
01:29:54,440 --> 01:29:58,840
because not everyone is going to be able to access the infrastructure

1928
01:30:00,120 --> 01:30:05,720
because they are not enough infrastructure or because they're regularly part.

1929
01:30:05,800 --> 01:30:08,520
And the Aurora is going to be crucial.

1930
01:30:08,520 --> 01:30:11,960
They got the role of open source and people that's...

1931
01:30:11,960 --> 01:30:15,960
The open models, open systems, all those things are going to be very important.

1932
01:30:15,960 --> 01:30:18,920
I think that's one thing that the 2027 article got right.

1933
01:30:18,920 --> 01:30:23,960
Like they are predicting an almost like an ideological arms race in some sense,

1934
01:30:23,960 --> 01:30:24,520
I would say.

1935
01:30:24,520 --> 01:30:27,720
I mean, they phrase it as like a technological political one,

1936
01:30:27,720 --> 01:30:32,680
but it's like basically whoever controls the models controls the future kind of scenario.

1937
01:30:32,680 --> 01:30:35,000
And it's like US versus China, you know.

1938
01:30:36,120 --> 01:30:40,600
And it's already clear now that, you know, like US models have...

1939
01:30:42,040 --> 01:30:44,120
I mean, that's a gross over generalization,

1940
01:30:44,120 --> 01:30:47,240
but let's say US models have Western values imbued,

1941
01:30:47,240 --> 01:30:51,080
and Chinese models have the values of the Communist Party.

1942
01:30:52,440 --> 01:30:53,240
Embedded.

1943
01:30:53,240 --> 01:30:58,920
And it's basically if you want to, you know, move the needle in terms of

1944
01:31:00,040 --> 01:31:02,360
the ideological stance of the world, it's like,

1945
01:31:02,920 --> 01:31:06,920
if your model is the default model, then you can really move the needle.

1946
01:31:06,920 --> 01:31:07,400
Oh, yeah.

1947
01:31:07,400 --> 01:31:12,040
And they predicted that this is going to play out over the next couple of years.

1948
01:31:12,040 --> 01:31:13,320
And I do agree with that.

1949
01:31:13,320 --> 01:31:16,360
Like I think this is already happening and this is going to happen.

1950
01:31:16,360 --> 01:31:20,680
To your case, you know, like if you have an idea for a new business or what have you

1951
01:31:20,680 --> 01:31:24,440
and you prompt the LLM to help you set it up and all those kind of things.

1952
01:31:24,440 --> 01:31:28,680
Like if it's the CCP model, then it's going to be a slightly different business

1953
01:31:28,680 --> 01:31:30,200
than if it's another model and so on.

1954
01:31:30,200 --> 01:31:32,440
So I think all of that is already happening.

1955
01:31:32,440 --> 01:31:34,600
So it's just like, I think that, yeah, I don't,

1956
01:31:34,600 --> 01:31:36,040
I'm not sure what to make of all of it.

1957
01:31:36,040 --> 01:31:39,400
You know, I'm very close to just moving to, I don't know,

1958
01:31:39,400 --> 01:31:43,160
to a mountain hut or to a cave or something, just touching grass forever.

1959
01:31:44,360 --> 01:31:46,760
Yeah, not touching a keyboard anymore.

1960
01:31:46,760 --> 01:31:51,320
I mean, that's why I like, I move away from keyboards now and just speak into microphones.

1961
01:31:53,320 --> 01:31:55,080
That's a step forward.

1962
01:31:55,080 --> 01:31:55,960
Yeah, yeah, yeah.

1963
01:31:55,960 --> 01:31:59,160
It's a one step away from technology.

1964
01:31:59,240 --> 01:32:01,640
But in all seriousness, like what me,

1965
01:32:01,640 --> 01:32:06,200
what was most fascinating to me the last couple of weeks,

1966
01:32:06,200 --> 01:32:08,040
like really exploring all that,

1967
01:32:08,040 --> 01:32:10,920
or yeah, the last two, three months now or so,

1968
01:32:10,920 --> 01:32:16,200
is how, how dialogical everything with LLM is.

1969
01:32:16,200 --> 01:32:19,720
Like you really enter into a dialogue with the technology itself.

1970
01:32:19,720 --> 01:32:26,920
Like you, the natural language interface is just like a constant back and forth

1971
01:32:27,480 --> 01:32:29,160
prompting and reading and correcting.

1972
01:32:29,160 --> 01:32:31,560
And you are actually in the dialogue,

1973
01:32:31,560 --> 01:32:36,360
like if you're, like, it doesn't matter if you're using cursor or a rule or what have you,

1974
01:32:37,640 --> 01:32:40,760
you're in a constant dialogue with your IDE, for example.

1975
01:32:40,760 --> 01:32:41,320
Yeah.

1976
01:32:41,320 --> 01:32:44,920
And that is very strange and it is very new in some sense.

1977
01:32:44,920 --> 01:32:47,240
It is very, very different.

1978
01:32:47,240 --> 01:32:51,000
And that's also what I liked about, you know,

1979
01:32:51,000 --> 01:32:52,920
like kind of us exploring all that together,

1980
01:32:52,920 --> 01:32:55,720
because it was kind of the constant dialogue about all those things

1981
01:32:55,720 --> 01:32:57,960
and all the topics you touched on as well.

1982
01:32:57,960 --> 01:33:02,920
And kind of, I think that's the only thing we have to navigate anything.

1983
01:33:02,920 --> 01:33:05,560
It's like constant dialogue and different constant exploration.

1984
01:33:05,560 --> 01:33:06,200
Yeah.

1985
01:33:06,200 --> 01:33:10,280
And I think that also works to spur the overlap between, you know,

1986
01:33:10,280 --> 01:33:13,800
the natural language that we use right now to communicate.

1987
01:33:14,440 --> 01:33:17,080
And the natural language that we use to prompt.

1988
01:33:17,080 --> 01:33:20,040
There is a big overlap between these two topics.

1989
01:33:20,040 --> 01:33:22,760
And I think we were talking about that the other day,

1990
01:33:22,760 --> 01:33:26,680
but I noticed that by prompting LLMs,

1991
01:33:26,680 --> 01:33:29,160
I'm being very clear and consistent.

1992
01:33:29,160 --> 01:33:30,520
I'm thinking something.

1993
01:33:30,520 --> 01:33:31,000
Yeah, yeah.

1994
01:33:31,000 --> 01:33:35,880
And to put my ideas in order in the way that makes sense for a standard system,

1995
01:33:35,880 --> 01:33:37,240
that is kind of neutral.

1996
01:33:38,520 --> 01:33:43,640
It kind of helped me to formulate things in order to express it to other people and so on.

1997
01:33:45,640 --> 01:33:47,880
And there is something very interesting to spur,

1998
01:33:47,880 --> 01:33:51,240
like the overlap between the natural language interface.

1999
01:33:52,920 --> 01:33:56,600
Definitely, we are into something here because

2000
01:33:57,640 --> 01:34:03,160
the natural language interface is something that's going to prevail in the time.

2001
01:34:03,160 --> 01:34:03,560
Oh, yeah.

2002
01:34:04,760 --> 01:34:13,960
It's going to be like the same that from toes keyboards to, sorry,

2003
01:34:13,960 --> 01:34:16,920
to physical keyboards, to toes screens, you know.

2004
01:34:16,920 --> 01:34:18,280
It's going to be the same.

2005
01:34:19,240 --> 01:34:24,280
You had at the beginning one interface with physical keys that you have to,

2006
01:34:25,160 --> 01:34:31,560
I prefer that to be honest, but then we have the toes screens that are more intuitive than you can.

2007
01:34:31,560 --> 01:34:34,600
No, it's a massive, massive evolution in terms of interfaces.

2008
01:34:34,600 --> 01:34:38,040
It's also, I like what I don't remember now who said it.

2009
01:34:38,040 --> 01:34:46,040
Maybe it was hazard, kind of the next evolutionary step in terms of how we went from,

2010
01:34:46,760 --> 01:34:50,440
you know, like, assembly to higher level languages.

2011
01:34:50,440 --> 01:34:53,640
And now we're going, like, higher level languages, like C.

2012
01:34:56,680 --> 01:35:02,440
And now we're going from high level languages to natural language,

2013
01:35:02,440 --> 01:35:05,480
which is natural language and natural concepts, basically,

2014
01:35:05,480 --> 01:35:09,320
which is, of course, a layer above and then it gets translated to, you know,

2015
01:35:09,320 --> 01:35:10,920
like, TypeScript, most likely.

2016
01:35:12,600 --> 01:35:13,640
Some form of JavaScript.

2017
01:35:14,600 --> 01:35:17,480
Like, if you don't define the language, it will just default to JavaScript.

2018
01:35:17,480 --> 01:35:18,440
It's really funny.

2019
01:35:18,440 --> 01:35:23,080
It's just like, okay, everything I read, like, 90% of the code I read is JavaScript.

2020
01:35:23,080 --> 01:35:24,440
So I'm just going to use JavaScript.

2021
01:35:24,440 --> 01:35:25,800
Humans really like JavaScript.

2022
01:35:25,800 --> 01:35:27,400
That's basically what they are thinking.

2023
01:35:27,400 --> 01:35:28,760
Oh my God, I hate JavaScript.

2024
01:35:28,760 --> 01:35:30,280
The JavaScript was make for humans.

2025
01:35:31,720 --> 01:35:35,160
Anyway, now that I don't have to write or read it anymore, I can just vibe it.

2026
01:35:36,200 --> 01:35:37,160
I like JavaScript now.

2027
01:35:37,160 --> 01:35:37,640
It's okay.

2028
01:35:37,640 --> 01:35:38,200
It's okay.

2029
01:35:38,200 --> 01:35:43,160
I mean, we should change our natural language and use speaking HTML, CSS,

2030
01:35:43,240 --> 01:35:44,200
and JavaScript.

2031
01:35:44,200 --> 01:35:44,760
Yeah, exactly.

2032
01:35:45,640 --> 01:35:47,560
It would be a much better structure than me.

2033
01:35:47,560 --> 01:35:50,040
Yeah, that would work perfectly well.

2034
01:35:50,040 --> 01:35:51,640
Marka, let's see.

2035
01:35:51,640 --> 01:35:52,600
What do you say?

2036
01:35:52,600 --> 01:35:53,320
Choose?

2037
01:35:53,320 --> 01:35:53,640
Yes.

2038
01:35:53,640 --> 01:35:54,680
I'm pretty cooked.

2039
01:35:54,680 --> 01:35:55,080
Yeah.

2040
01:35:55,080 --> 01:35:55,560
All right.

2041
01:35:55,560 --> 01:35:56,440
Any final words?

2042
01:35:57,480 --> 01:35:58,120
Last words.

2043
01:35:58,120 --> 01:36:00,520
Any last words before Skynet takes over?

2044
01:36:00,520 --> 01:36:01,960
And we're all going to make it?

2045
01:36:01,960 --> 01:36:02,840
We're all going to make it?

2046
01:36:03,480 --> 01:36:04,440
That's fantastic.

2047
01:36:06,040 --> 01:36:07,240
Those are great final words.

2048
01:36:08,440 --> 01:36:09,960
Let's see if we're still recording.

2049
01:36:09,960 --> 01:36:11,160
Oh, yeah, we're lucky.

2050
01:36:11,160 --> 01:36:11,880
Nice.