1
00:00:00,060 --> 00:00:10,000
Speaker 1: [upbeat music]

2
00:00:10,080 --> 00:00:13,760
Speaker 1: Hello, and welcome to a new episode
of "The CTO Show with Mehmet." My name is

3
00:00:13,820 --> 00:00:17,800
Speaker 1: Mehmet, and as you know, in each
episode I discuss different topics about
emerging

4
00:00:17,880 --> 00:00:23,440
Speaker 1: technologies from AI, digital
transformation, cybersecurity. And also I

5
00:00:23,500 --> 00:00:28,080
Speaker 1: sometimes have guests with me who are
subject matter experts in one of the domains I

6
00:00:28,140 --> 00:00:32,340
Speaker 1: usually talk about. And today I'm
very pleased to have with me Steven Schrembeck,

7
00:00:32,420 --> 00:00:37,580
Speaker 1: who's joining me from the United
States, from Georgia. Steven, thank you very
much

8
00:00:37,640 --> 00:00:41,760
Speaker 1: for being on the show today. I will
keep it to you to introduce yourself, what you

9
00:00:41,800 --> 00:00:43,760
Speaker 1: do, and what you are up to.

10
00:00:45,420 --> 00:00:50,920
Speaker 2: Hey, Mehmet. So thanks for having me
on. So I do three things that are of relevance.

11
00:00:51,300 --> 00:00:56,780
Speaker 2: One, so I'm the founder of a startup
called Impossible Labs, so it builds AI stuff.

12
00:00:57,240 --> 00:01:01,960
Speaker 2: It sounds very fancy, but I'm
basically a solopreneur plus AI, [chuckles] plus

13
00:01:02,000 --> 00:01:07,760
Speaker 2: contractors, and I, I really like it
that way. Uh, beyond that, I make videos on a

14
00:01:07,800 --> 00:01:12,340
Speaker 2: channel called Seeking Minima, so
that's what brought us here today. That channel

15
00:01:12,380 --> 00:01:17,560
Speaker 2: has one purpose, and that is to help
people use technology instead of being used by

16
00:01:17,640 --> 00:01:22,320
Speaker 2: it, and that's sort of my goal. And
then the, the last part is I write stories,

17
00:01:22,820 --> 00:01:23,780
Speaker 2: science fiction and fantasy.

18
00:01:24,680 --> 00:01:27,720
Speaker 1: Oh, wow. I love this combination,
really.

19
00:01:28,160 --> 00:01:28,240
Speaker 2: Mm-hmm.

20
00:01:28,380 --> 00:01:34,080
Speaker 1: Now, uh, I know Steven also writing
like you, you do software development as well,

21
00:01:34,180 --> 00:01:34,440
Speaker 1: right?

22
00:01:35,680 --> 00:01:38,920
Speaker 2: That's right. Yeah. I didn't mention
the one thing that like pays the bills, right?

23
00:01:39,520 --> 00:01:44,560
Speaker 2: Um, so I guess AI does pay the bills.
Um, but yeah, so I've been a software

24
00:01:44,600 --> 00:01:50,080
Speaker 2: developer for 12 years, something
like that. Um, big corporate software developer.

25
00:01:50,500 --> 00:01:53,400
Speaker 2: Uh, I made the rounds. I worked at
Amazon for a while, uh-

26
00:01:53,460 --> 00:01:53,461
Speaker 1: Okay

27
00:01:53,460 --> 00:01:59,240
Speaker 2: ... worked at smaller companies. So
it's been good to me, uh, but it's not,

28
00:01:59,760 --> 00:02:03,900
Speaker 2: it's not as exciting as, um, cutting
edge tech, which is always where my heart has

29
00:02:03,940 --> 00:02:04,120
Speaker 2: been.

30
00:02:05,820 --> 00:02:11,680
Speaker 1: Uh, that's great. So Steven, just wh-
out of curiosity, like you have a mix of,

31
00:02:12,100 --> 00:02:17,940
Speaker 1: you know, multiple things at the same
time. So what I would say

32
00:02:18,220 --> 00:02:22,180
Speaker 1: drove you to choose this path, uh,
for your career?

33
00:02:25,180 --> 00:02:25,400
Speaker 2: So

34
00:02:27,680 --> 00:02:31,880
Speaker 2: I've moved through different kinds of
software development and first of all, I can't

35
00:02:31,920 --> 00:02:35,540
Speaker 2: help myself. That's the easy answer,
right? [chuckles] It's, I can't help but be

36
00:02:35,600 --> 00:02:39,100
Speaker 2: interested in things. Writing is
something I haven't been able to put down for a

37
00:02:39,140 --> 00:02:43,620
Speaker 2: long time, so that's just like
something that won't leave me alone. Um, and
same

38
00:02:43,640 --> 00:02:48,920
Speaker 2: with cutting edge tech, right? Uh, I
moved into software development because I liked

39
00:02:49,440 --> 00:02:54,760
Speaker 2: building things. I'm not one of those
people that loves code for code's sake. I, I

40
00:02:54,800 --> 00:02:59,180
Speaker 2: like it as a tool, and while I am
fascinated with building things and
understanding

41
00:02:59,220 --> 00:03:05,100
Speaker 2: how they work, I'm more interested in
the opportunity to solve cool problems and

42
00:03:05,240 --> 00:03:09,040
Speaker 2: help people. And software was just
the easiest way to do that 'cause as you know,

43
00:03:09,120 --> 00:03:13,280
Speaker 2: it's easy to set up, tear down, like
there's, there's no sandbox like software. Um,

44
00:03:13,360 --> 00:03:18,080
Speaker 2: so I think that's really what drew me
in. And then AI is just... Honestly, it felt

45
00:03:18,100 --> 00:03:23,660
Speaker 2: like modern day magic. I started
building models in 2016, uh, when deep learning
was

46
00:03:23,700 --> 00:03:28,380
Speaker 2: having a first or second renaissance,
depending on how you look at it. And I knew

47
00:03:28,420 --> 00:03:33,320
Speaker 2: that this was computer magic
[chuckles] and I wanted to know how to do it.
Uh, so

48
00:03:33,520 --> 00:03:34,680
Speaker 2: that's what drew me into AI.

49
00:03:36,500 --> 00:03:42,420
Speaker 1: That's very, very cool I would say.
Now to- l- I know this is the hot topic. It's

50
00:03:42,480 --> 00:03:48,100
Speaker 1: on everyone's mind, right? So it's
AI. Um, how do you

51
00:03:48,140 --> 00:03:53,900
Speaker 1: perceive it from not only a developer
perspective, a solopreneur perspective,

52
00:03:54,400 --> 00:03:59,420
Speaker 1: how do you look to AI? Do you see, do
you see it as, you know, the technology that

53
00:03:59,540 --> 00:04:05,280
Speaker 1: will, you know, make really people's
life easy or do you see it

54
00:04:05,680 --> 00:04:11,520
Speaker 1: more as a, just a cool technology
that we can do cool stuff with it?

55
00:04:12,480 --> 00:04:15,980
Speaker 1: How, how do you perceive really the
technology? Because there are a lot-- And the

56
00:04:16,019 --> 00:04:20,680
Speaker 1: reason I'm asking you, Steven, this
question, there are a lot of debates recently

57
00:04:20,740 --> 00:04:26,120
Speaker 1: with all the hype that happened after
ChatGPT and all these technologies. What's

58
00:04:26,160 --> 00:04:30,920
Speaker 1: next? What's going to happen in the
world? So from a developer perspective, a

59
00:04:30,960 --> 00:04:34,600
Speaker 1: solopreneur perspective, what you can
tell us about this?

60
00:04:37,940 --> 00:04:42,940
Speaker 2: So a lot of the time the hype isn't
real. So I also, you know, I'm very interested

61
00:04:42,980 --> 00:04:47,380
Speaker 2: in Web3 mostly again for like
coordination technologies. I'm terrible at, uh,

62
00:04:47,400 --> 00:04:51,220
Speaker 2: [chuckles] I'm terrible at riding the
hype wave. I'm always interested in the tech

63
00:04:51,240 --> 00:04:56,960
Speaker 2: and the values. Um, so for AI, this
time the hype is kind of real, and it was last

64
00:04:56,980 --> 00:05:01,180
Speaker 2: time too. Uh, if you'll recall, I
don't know what, four or five years ago, the
last

65
00:05:01,200 --> 00:05:04,920
Speaker 2: time deep learning really took off it
was for vision networks, and this is really

66
00:05:04,960 --> 00:05:09,520
Speaker 2: where the killer use case was. And
the hype was real, it just wasn't evenly

67
00:05:09,560 --> 00:05:14,520
Speaker 2: distributed. Now, as, as you know,
diffusion networks for generative AI and large

68
00:05:14,560 --> 00:05:19,960
Speaker 2: language models, uh, are the latest
hotness and the hype really is real again. Um,

69
00:05:21,180 --> 00:05:26,800
Speaker 2: I think the reason that we might not
see another AI winter between here and

70
00:05:26,980 --> 00:05:32,860
Speaker 2: AI really doing very impactful things
to society is because

71
00:05:32,900 --> 00:05:37,580
Speaker 2: we have a lot of Lego blocks that
we've built up. If you followed AI development
for

72
00:05:37,640 --> 00:05:43,060
Speaker 2: the past, you know, decade or so, it
keeps increasing. But what has happened is that

73
00:05:43,100 --> 00:05:47,120
Speaker 2: everyone was increasing in their own
silos. So, you know, nat- natural language

74
00:05:47,200 --> 00:05:52,460
Speaker 2: processing was getting better. Um,
time series analysis was getting better. Just

75
00:05:52,500 --> 00:05:56,080
Speaker 2: transformer networks, all, all the
vision networks, all of these things were

76
00:05:56,120 --> 00:06:01,200
Speaker 2: improving. Diffusion, autoencoders,
all these things were improving separately. And

77
00:06:01,240 --> 00:06:06,580
Speaker 2: what happened is people from one
domain connected a piece to another domain. So
in

78
00:06:06,620 --> 00:06:11,520
Speaker 2: this case, it was reinforcement
learning plus natural language processing. Um,
and

79
00:06:11,620 --> 00:06:15,680
Speaker 2: plus-- And that was basically you
connect two pieces from two verticals that have

80
00:06:15,700 --> 00:06:20,740
Speaker 2: been advancing for a long time, and
the result was exponential increase. It could do

81
00:06:20,760 --> 00:06:25,100
Speaker 2: things. The difference was we have
generative networks. We've had GANs for a long

82
00:06:25,200 --> 00:06:30,000
Speaker 2: time, but they're sort of obtuse and
really hard to train. The difference is we

83
00:06:30,080 --> 00:06:36,000
Speaker 2: needed, we needed an interface that
understood what we meant. That was the big,

84
00:06:36,060 --> 00:06:41,380
Speaker 2: like, unlock here, is that we could
always sort of get networks to do really good

85
00:06:41,420 --> 00:06:45,440
Speaker 2: things narrowly, but we couldn't
interface with them in a, in a very good way. If

86
00:06:45,480 --> 00:06:50,740
Speaker 2: they know something like, um, like a
large language model does, how do you get the

87
00:06:50,780 --> 00:06:55,540
Speaker 2: information out of it? And I think
that natural language processing was an
interface

88
00:06:55,660 --> 00:07:01,460
Speaker 2: moment. So the combination of all
these individual advancements is

89
00:07:01,500 --> 00:07:06,700
Speaker 2: really why you're seeing huge changes
now, and the amount of individual advancement

90
00:07:06,740 --> 00:07:10,900
Speaker 2: has not stopped. So I would be very
surprised if there were not more verticals to

91
00:07:10,920 --> 00:07:16,740
Speaker 2: combine and again, create another
ridiculous pivotal moment every six

92
00:07:16,800 --> 00:07:20,880
Speaker 2: months or so. But you might see lulls
in the meantime. So I'd say the hype is real,

93
00:07:22,080 --> 00:07:26,280
Speaker 2: sort of, right? [chuckles] People
always tend to overhype no matter what you do.

94
00:07:27,740 --> 00:07:33,380
Speaker 1: Um, actually, Steven, you said
something important which I discussed couple of,
uh,

95
00:07:33,500 --> 00:07:39,040
Speaker 1: episodes back. I, I go solo sometime
and I said like I'm, I'm an old, not very old

96
00:07:39,120 --> 00:07:44,440
Speaker 1: guy, but I'm an old guy enough to
remember all the hypes in history and I don't

97
00:07:44,500 --> 00:07:50,380
Speaker 1: remember a moment, you know, similar
to what we are seeing today. Because I was like

98
00:07:51,120 --> 00:07:55,900
Speaker 1: maybe 10, 11 years old when the first
time I heard about the internet. You know, I

99
00:07:55,940 --> 00:08:00,020
Speaker 1: was 14 years old when I tried the
internet and then, you know, later on the

100
00:08:00,620 --> 00:08:06,220
Speaker 1: smartphone, the Web2. Um, yeah, like
there was

101
00:08:06,280 --> 00:08:10,800
Speaker 1: sometimes exaggeration, sometimes,
you know, things going, but you know, the, the

102
00:08:10,920 --> 00:08:16,380
Speaker 1: cycle used to take very long until we
see the next thing. But with AI, I mean

103
00:08:16,440 --> 00:08:21,100
Speaker 1: generative AI and the NLPs and, you
know, these all models that we are seeing today,

104
00:08:21,240 --> 00:08:25,580
Speaker 1: I think we, we, you know, as you
said, I like the word you said, an AI we get--
we

105
00:08:25,600 --> 00:08:31,440
Speaker 1: will not see an AI winter. I love
this expression and I think we are going very

106
00:08:31,560 --> 00:08:37,300
Speaker 1: fast. Now from a, I would say
developer perspective, right? What do you think
would

107
00:08:37,360 --> 00:08:43,240
Speaker 1: be the best applications that let me
make it very general

108
00:08:43,280 --> 00:08:46,100
Speaker 1: that everyone can benefit on other
than, you know, the,

109
00:08:47,700 --> 00:08:53,480
Speaker 1: I would say the sometimes very simple
things we see, write me an email, write me

110
00:08:53,680 --> 00:08:58,540
Speaker 1: this. So I believe like from your
perspective, Steven, you see a bigger picture.
Can

111
00:08:58,580 --> 00:09:00,540
Speaker 1: you share that with us from your
perspective?

112
00:09:02,300 --> 00:09:02,580
Speaker 2: Yeah.

113
00:09:04,160 --> 00:09:09,640
Speaker 2: I think another reason why this time
it's different is we have a very general model,

114
00:09:09,880 --> 00:09:14,380
Speaker 2: a shockingly general model, meaning
it can, it can reason about things, right?

115
00:09:14,460 --> 00:09:17,840
Speaker 2: Whether it doesn't really matter
what's happening under the hood, all that
matters

116
00:09:17,900 --> 00:09:21,880
Speaker 2: is the input output results in
something that looks like commodified
intelligence.

117
00:09:21,940 --> 00:09:26,740
Speaker 2: It's a brain in a box. You can spin
it up. Now it's not super smart. It's like maybe

118
00:09:26,840 --> 00:09:31,420
Speaker 2: average adult smart, and then in a
couple domains it's exceptionally smart, like
law

119
00:09:31,480 --> 00:09:37,460
Speaker 2: or some solving SAT problems. But
what matters is that you have common

120
00:09:37,560 --> 00:09:42,120
Speaker 2: sense reasoning in a box. That's the
kind of thing that honestly I expected to take

121
00:09:42,180 --> 00:09:45,600
Speaker 2: a lot longer and I think a lot of
people did. I did not see this one coming next.

122
00:09:46,040 --> 00:09:49,520
Speaker 2: There was a number of hurdles to
artificial general intelligence and I didn't
think

123
00:09:49,560 --> 00:09:55,460
Speaker 2: this one was gonna get knocked over
first or next anyway. So really when you think

124
00:09:55,540 --> 00:10:00,660
Speaker 2: about how can I use this, what could
you ask somebody that you could pay minimum

125
00:10:00,720 --> 00:10:04,980
Speaker 2: wage or just somebody with no
training of average intelligence to do for you
on a

126
00:10:05,000 --> 00:10:10,980
Speaker 2: computer? Now you can do that for
instead of $15 an hour, depending on where

127
00:10:11,020 --> 00:10:16,639
Speaker 2: you live, you can do that for 20
cents and you can do it a thousand times

128
00:10:16,900 --> 00:10:22,840
Speaker 2: faster and you can have 100,000 of
them working. So what can you do with that?

129
00:10:23,060 --> 00:10:28,580
Speaker 2: Well, any information processing is
if your job is information processing,

130
00:10:30,200 --> 00:10:34,080
Speaker 2: that's AI job now. Um, and you're
gonna do something else. So it's kind of
hollowing

131
00:10:34,140 --> 00:10:39,320
Speaker 2: out the middle. It's hollowing out
people who work on-- If your job is to turn one

132
00:10:39,400 --> 00:10:44,320
Speaker 2: piece of information into another
piece of information, if there is a broad skill

133
00:10:44,420 --> 00:10:48,980
Speaker 2: set and data set for what you do,
you're gonna be replaced pretty quickly, um,

134
00:10:49,600 --> 00:10:55,360
Speaker 2: depending on the economic output of
what it's worth to replace you. Except if you're

135
00:10:55,380 --> 00:11:00,540
Speaker 2: a specialist. There is, is very hard
to replace specialists. So I know a person

136
00:11:00,620 --> 00:11:06,140
Speaker 2: whose job is to audit, uh, oil and
gas refinery, uh, machinery for safety.

137
00:11:06,840 --> 00:11:11,600
Speaker 2: And even though his job is basically
to analyze data and spit out, okay, safe, not

138
00:11:11,660 --> 00:11:16,520
Speaker 2: safe, do maintenance, don't do
maintenance, replace, he's not gonna be replaced
for

139
00:11:16,560 --> 00:11:22,480
Speaker 2: a very long time because it's such a
specialized knowledge set that he's fine. So

140
00:11:22,940 --> 00:11:28,780
Speaker 2: if your job is pretty generic junior
coder with no specialization or pretty

141
00:11:28,860 --> 00:11:32,420
Speaker 2: generic junior designer who just does
websites,

142
00:11:34,000 --> 00:11:38,540
Speaker 2: you should consider using other
things. Now I think your question was to ask how
can

143
00:11:38,580 --> 00:11:41,480
Speaker 2: people use this? That was sort of
like the negative side-

144
00:11:41,540 --> 00:11:41,660
Speaker 1: Yeah

145
00:11:41,760 --> 00:11:46,400
Speaker 2: ... which gets a lot of attention.
But the positive side is true too. This is

146
00:11:46,440 --> 00:11:51,460
Speaker 2: commodified intelligence in a box. If
something, if there's enough data, you can

147
00:11:51,500 --> 00:11:53,980
Speaker 2: generate it. But that's the very
general thing.

148
00:11:56,160 --> 00:12:01,700
Speaker 2: I can tell you that what I'm using it
for is to remove, is to really take

149
00:12:01,740 --> 00:12:07,632
Speaker 2: control of my sort of- To fight back
against the attention economy, I guess is the

150
00:12:07,672 --> 00:12:11,872
Speaker 2: right way of putting it. So just for
instance, yesterday-- Actually this morning and

151
00:12:11,912 --> 00:12:17,892
Speaker 2: last night, I wrote a tool that uses
OpenAI's Whisper to transcribe podcasts. So

152
00:12:17,912 --> 00:12:21,792
Speaker 2: it takes from audio to text to give
you a transcript, including speaker attribution,

153
00:12:21,892 --> 00:12:26,832
Speaker 2: so who's talking. And then it takes
that, pumps it into ChatGPT-4 through the API,

154
00:12:26,912 --> 00:12:32,252
Speaker 2: [chuckles] and then I run it through
a prompt where I tell it basically, "All right.

155
00:12:32,392 --> 00:12:35,692
Speaker 2: Here's the questions I'm gonna ask.
Here are the things I need to know about this

156
00:12:35,732 --> 00:12:39,552
Speaker 2: podcast. Here are the viewpoints I
wanna see." And then it just curates this
report.

157
00:12:40,212 --> 00:12:43,532
Speaker 2: So there are certain podcasts, let's
say like macro investing, right? I don't really

158
00:12:43,572 --> 00:12:47,532
Speaker 2: care about this too much. I wanna
know roughly, you know, where the markets are. I

159
00:12:47,552 --> 00:12:52,232
Speaker 2: don't wanna pay attention to it. News
is-- makes you feel bad anyway. But now, with

160
00:12:52,272 --> 00:12:56,532
Speaker 2: the push of a button, here's
everything I need to know distilled down into a
s- tiny

161
00:12:56,592 --> 00:13:00,312
Speaker 2: amount, and I can curate that for
many podcast feeds. And this is just stuff I'm

162
00:13:00,332 --> 00:13:03,392
Speaker 2: making. So I am using it to sort of

163
00:13:04,932 --> 00:13:10,672
Speaker 2: get more information and learn more
than I could before, and to underst- to learn

164
00:13:10,772 --> 00:13:15,872
Speaker 2: quickly and to synthesize
information. I'm using it like intelligence in a
box.

165
00:13:15,892 --> 00:13:17,292
Speaker 2: "Hey, read this for me."

166
00:13:18,152 --> 00:13:18,292
Speaker 1: Mm-hmm.

167
00:13:18,532 --> 00:13:23,372
Speaker 2: "Hey, distill this down and..." But
only in the places where my goal is just

168
00:13:23,412 --> 00:13:27,732
Speaker 2: information. So if you wanna be
matrix mainlined, just plug me in. Like, I just

169
00:13:27,772 --> 00:13:32,832
Speaker 2: wanna know how to do this. This is a,
a killer use case right now for anybody. You

170
00:13:32,852 --> 00:13:36,972
Speaker 2: don't have to know how to program.
Just interfacing with the, the chat clients is

171
00:13:37,012 --> 00:13:42,592
Speaker 2: enough. Uh, you could be learning
really, really quickly right now. And so this is

172
00:13:42,712 --> 00:13:44,912
Speaker 2: the sort of-- This is the trade.

173
00:13:45,592 --> 00:13:45,632
Speaker 1: Yeah.

174
00:13:45,652 --> 00:13:49,472
Speaker 2: So you have an opportunity where AI
is really, really useful at helping you learn

175
00:13:49,812 --> 00:13:54,432
Speaker 2: just about anything that makes you
very marketable right now. Or you can ignore it

176
00:13:54,572 --> 00:13:59,812
Speaker 2: [chuckles] and sit around and wait
for it to-- your job to be slowly eaten away.
Um,

177
00:14:00,052 --> 00:14:04,932
Speaker 2: so it's like you can ha- you can
really stand out right now if you're willing to
do

178
00:14:04,972 --> 00:14:08,752
Speaker 2: the work and learn and try these
things out, or you can, you know, wait on the

179
00:14:08,792 --> 00:14:09,212
Speaker 2: sidelines.

180
00:14:11,012 --> 00:14:16,492
Speaker 1: Wow. You know, like what you are
trying to build is, uh, something similar. Uh, I

181
00:14:16,552 --> 00:14:22,512
Speaker 1: mean, not exactly the same idea, but
I wanted to... I rely on

182
00:14:22,552 --> 00:14:28,472
Speaker 1: preparing this podcast on some news
feeds. So, and one of the use cases I thought is

183
00:14:29,292 --> 00:14:34,672
Speaker 1: let ChatGPT or whatever API read
these feeds and just select which ones are

184
00:14:34,712 --> 00:14:40,372
Speaker 1: important so I can choose a topic,
for example. Now, you mentioned

185
00:14:40,732 --> 00:14:46,492
Speaker 1: the word prompt several times, and
there's a hype, if we can call it, about prompt

186
00:14:46,532 --> 00:14:51,832
Speaker 1: engineers, right? So prompt engineers
and, you know, from your perspective, and

187
00:14:51,892 --> 00:14:56,672
Speaker 1: let's discuss it both, you know, from
a real software development perspective and

188
00:14:56,832 --> 00:15:02,472
Speaker 1: plus, you know, a realistic use case
perspective. How important is the prompt

189
00:15:02,812 --> 00:15:05,392
Speaker 1: when interacting with models like
ChatGPT?

190
00:15:08,732 --> 00:15:13,592
Speaker 2: Ideally, over time, the prompt
becomes less important. If you have a good
natural

191
00:15:13,652 --> 00:15:18,292
Speaker 2: language model, the goal is for it to
be a better interface. So I'll, I'll just take

192
00:15:18,332 --> 00:15:20,712
Speaker 2: a step back and do a little technical
explanation that I-

193
00:15:21,012 --> 00:15:21,332
Speaker 1: Yes, please

194
00:15:21,352 --> 00:15:24,532
Speaker 2: ... hopefully doesn't put people to
sleep, but I think is very useful for

195
00:15:24,572 --> 00:15:30,352
Speaker 2: understanding how these things work.
So at the center of

196
00:15:30,872 --> 00:15:35,932
Speaker 2: these large language models, and even
stuff like diffusion models, um, is something

197
00:15:36,012 --> 00:15:40,872
Speaker 2: that researchers call latent space,
which in my opinion, they just use a technical

198
00:15:40,932 --> 00:15:44,932
Speaker 2: sounding term to describe something
that's very complicated. I like to think of it

199
00:15:44,992 --> 00:15:49,912
Speaker 2: as just this big information soup.
It's encoded in a sort of machine language. It's

200
00:15:49,952 --> 00:15:53,952
Speaker 2: just a representation that the model
has learned about the wor- the world in the

201
00:15:54,032 --> 00:15:58,292
Speaker 2: most dense format, right? This is the
center of everything it knows. This is where

202
00:15:58,352 --> 00:16:03,692
Speaker 2: it's recorded all the information it
knows, all the patterns it needs to find. You

203
00:16:03,712 --> 00:16:08,752
Speaker 2: can think of it, it's like its
repository of knowledge, but it's not in words.
It's

204
00:16:08,832 --> 00:16:14,192
Speaker 2: in, technically in embeddings. Uh,
it's just hyper-compressed information.

205
00:16:14,992 --> 00:16:20,712
Speaker 2: So it knows a lot of stuff about a
lot of stuff. You know, the, the OpenAI models

206
00:16:20,752 --> 00:16:26,432
Speaker 2: have read a bulk of the public
internet, so they know a lot of

207
00:16:26,472 --> 00:16:30,892
Speaker 2: things. In fact, they know a lot more
than we can pull out of them. And so a lot of

208
00:16:30,932 --> 00:16:36,312
Speaker 2: the trick is how do we get out-- I
know you can do this. How do I get you to know

209
00:16:36,332 --> 00:16:40,652
Speaker 2: what I mean and so that I can
synthesize, either do these operations, write
this

210
00:16:40,692 --> 00:16:42,712
Speaker 2: code, perform these tasks, help me-

211
00:16:42,752 --> 00:16:42,932
Speaker 1: Mm-hmm

212
00:16:42,952 --> 00:16:48,112
Speaker 2: ... read this thing, understand what
I'm trying to get? So that interface gap

213
00:16:48,332 --> 00:16:52,472
Speaker 2: between I know you have this
information in your latent space, in this ocean
of

214
00:16:52,512 --> 00:16:57,732
Speaker 2: information, I know you can do this.
I'm trying to get you to focus only on what I

215
00:16:57,772 --> 00:17:02,032
Speaker 2: want you to do, and there's some
misalignment there. So there's a gap in the

216
00:17:02,092 --> 00:17:05,612
Speaker 2: interface, which is do you know what
I mean, and can you-

217
00:17:05,692 --> 00:17:05,712
Speaker 1: Mm-hmm

218
00:17:05,712 --> 00:17:10,112
Speaker 2: ... sort of pull it out of what you
actually know how to do? So that's what prompt

219
00:17:10,172 --> 00:17:15,492
Speaker 2: engineering right now solves for, is
can you shrink that gap to be better at

220
00:17:15,792 --> 00:17:20,952
Speaker 2: synthesizing what you want from these
large language models? So right now it is

221
00:17:21,032 --> 00:17:27,032
Speaker 2: useful. I-- It will probably be
useful for a while, but in theory, as

222
00:17:27,052 --> 00:17:31,432
Speaker 2: things get better, that interface
should get better, and the models will get
better

223
00:17:31,452 --> 00:17:35,692
Speaker 2: and better at understanding what you
want so that you don't have to use all these

224
00:17:35,732 --> 00:17:39,412
Speaker 2: tricks of the trade. So I wouldn't
say there's a super long shelf life on this

225
00:17:39,452 --> 00:17:41,492
Speaker 2: skill, but I could be wrong.

226
00:17:43,472 --> 00:17:48,332
Speaker 1: And that could be [chuckles] -- You
know, like, because yesterday, when actually in

227
00:17:48,372 --> 00:17:54,152
Speaker 1: the, during the weekend, one of the
things I was trying to do is to let the

228
00:17:54,212 --> 00:17:59,852
Speaker 1: model tells me how best it prefers to
be interacted with.

229
00:18:00,572 --> 00:18:00,852
Speaker 1: And-

230
00:18:01,052 --> 00:18:01,332
Speaker 2: Got it

231
00:18:01,392 --> 00:18:07,292
Speaker 1: ... uh, I spent some time until, and
actually you, people might laugh, but

232
00:18:08,032 --> 00:18:13,268
Speaker 1: I told ChatGPT after I get the answer
I wanted, I said, "Okay." Can we write an

233
00:18:13,328 --> 00:18:18,948
Speaker 1: e-book about the... And put some
examples, and it did. So as per ChatGPT,

234
00:18:19,708 --> 00:18:24,448
Speaker 1: it prefers to have the following: a
context, specificity,

235
00:18:24,988 --> 00:18:30,988
Speaker 1: instruction, purpose or goal, and
limitation. So this is what I was told by

236
00:18:31,028 --> 00:18:35,848
Speaker 1: ChatGPT, and it gives me an example,
by the way. If you ask me this way, exactly as

237
00:18:35,888 --> 00:18:40,528
Speaker 1: you said, Steven, I know that I
should search exactly in this. You know, I don't

238
00:18:40,548 --> 00:18:44,488
Speaker 1: have to go search the whole language
now model, because I know exactly what I have

239
00:18:44,528 --> 00:18:48,828
Speaker 1: to go. So this is maybe where the
prompt engineering, but I believe, you know, I

240
00:18:48,868 --> 00:18:52,468
Speaker 1: have a little bit of technical
background and I know a little bit how these
models

241
00:18:52,528 --> 00:18:56,648
Speaker 1: work. I think the more they are
trained, the better they get, so even you will
not

242
00:18:56,728 --> 00:19:01,988
Speaker 1: need even, you know, an exact prompt
to get the information out of it, right? So

243
00:19:03,188 --> 00:19:08,468
Speaker 1: that's, that's very, very cool, I
would say. Now, here I want to ask you, now we

244
00:19:08,528 --> 00:19:14,188
Speaker 1: have this moment, I would say, what
do you think other technologies will get

245
00:19:14,248 --> 00:19:19,848
Speaker 1: combined or let's say, you know, used
together with the, you know, all this AI,

246
00:19:20,528 --> 00:19:25,748
Speaker 1: um, technologies? Like for example,
personally, I think automation with AI will play

247
00:19:25,788 --> 00:19:31,248
Speaker 1: a huge role. I started to see, you
know, someone talked about autonomous bots that

248
00:19:31,288 --> 00:19:36,068
Speaker 1: they can interact, you know, with
each other. What's your view on this?

249
00:19:37,408 --> 00:19:40,308
Speaker 2: All right. So the use... Just that's
like three good questions.

250
00:19:40,308 --> 00:19:41,008
Speaker 1: [chuckles]

251
00:19:41,088 --> 00:19:41,368
Speaker 2: Um,

252
00:19:43,048 --> 00:19:43,468
Speaker 2: so

253
00:19:45,388 --> 00:19:49,008
Speaker 2: I think that there's a lot of-- It
will be combined with everything. It's

254
00:19:49,028 --> 00:19:53,248
Speaker 2: intelligence, uh, so it, it gets
combined with everything. But I think the ones
that

255
00:19:53,268 --> 00:19:57,748
Speaker 2: are sort of juicy and next, I
actually don't think it's automation. Now, I
could be

256
00:19:57,768 --> 00:20:03,728
Speaker 2: wrong. There is a lot of value there.
So I think it's more like narrowly there are

257
00:20:03,748 --> 00:20:07,228
Speaker 2: new kinds of automation that were
just unlocked and those, yeah, that's
greenfield.

258
00:20:07,408 --> 00:20:11,508
Speaker 2: That'll be anything that required
like common sense reasoning. I think that those

259
00:20:11,548 --> 00:20:16,268
Speaker 2: things... There is-- Okay, automation
will move forward. The reason I don't think

260
00:20:16,308 --> 00:20:21,108
Speaker 2: that this is the automation moment, I
mean physical movement and manufacturing,

261
00:20:21,408 --> 00:20:27,308
Speaker 2: stuff in the real world, is because
physical stuff is hard. [chuckles] It's hard and

262
00:20:27,348 --> 00:20:33,168
Speaker 2: expensive and physics is finicky, and
moving stuff around in the real world requires

263
00:20:33,248 --> 00:20:38,808
Speaker 2: precision and rules and money and a
lot of testing that is a lot

264
00:20:38,968 --> 00:20:42,188
Speaker 2: slower than doing stuff in the
digital world. And this is something that
surprised

265
00:20:42,228 --> 00:20:46,448
Speaker 2: me, but I, as soon as I saw
diffusion, I don't know, what is that? A year
ago,

266
00:20:46,548 --> 00:20:51,248
Speaker 2: something like that. When stable
diffusion first came out, I knew immediately
that,

267
00:20:51,528 --> 00:20:57,248
Speaker 2: oh, I was wrong. It's not automation.
It's, it's coming for knowledge workers

268
00:20:57,308 --> 00:21:02,508
Speaker 2: first. It's coming for everything. So
that sounds very sinister, but like, because

269
00:21:02,548 --> 00:21:06,188
Speaker 2: that's where the data is. Not only is
that where the da- the data is, but the people

270
00:21:06,308 --> 00:21:07,148
Speaker 2: building these things,

271
00:21:08,688 --> 00:21:13,548
Speaker 2: this is what they understand, right?
They understand how to use computers, how to do

272
00:21:13,608 --> 00:21:19,148
Speaker 2: information jobs, how to program, how
to do design. So not only do they have the

273
00:21:19,168 --> 00:21:23,408
Speaker 2: data sets, but this is a dom- set of
domains they understand. So knowledge work will

274
00:21:23,428 --> 00:21:28,588
Speaker 2: be automated actually before robots.
So hilariously, it's the plumbers and the

275
00:21:28,828 --> 00:21:33,328
Speaker 2: mechanics and even the factory
workers of the world that will probably, uh, be

276
00:21:33,368 --> 00:21:38,288
Speaker 2: gainfully employed, you know, longer
than the sort of Silicon Valley types, which is

277
00:21:38,328 --> 00:21:44,148
Speaker 2: im- highly amusing. Um, but again,
it, that sounds sort of dark.

278
00:21:44,648 --> 00:21:48,508
Speaker 2: I think the other ways I would
combine this, I, I can just tell you what I'm
doing.

279
00:21:49,228 --> 00:21:53,648
Speaker 2: One of the products I'm most excited
about is called Intent. Um, so I'm trying to

280
00:21:53,688 --> 00:21:59,528
Speaker 2: build a system that literally watches
your screen. So this is a combination with

281
00:21:59,628 --> 00:22:04,588
Speaker 2: other data streams. So it's, it's
using vision nets to watch your screen and it

282
00:22:04,628 --> 00:22:09,088
Speaker 2: knows what you're doing, uh, all the
time. It's private, so it runs on just your

283
00:22:09,148 --> 00:22:09,528
Speaker 2: computer.

284
00:22:10,208 --> 00:22:10,368
Speaker 1: Mm-hmm.

285
00:22:10,428 --> 00:22:13,868
Speaker 2: And it just creates a stream of-- It
answers the question, what did I do all day,

286
00:22:14,588 --> 00:22:18,708
Speaker 2: right? It literally just keeps a log
because it can understand what you're up to.

287
00:22:19,088 --> 00:22:23,408
Speaker 2: Uh, right now you're doing an
interview and it can log moment by moment. So at
face

288
00:22:23,448 --> 00:22:28,628
Speaker 2: value it's just analytics, but the
real goal is to do executive control software.
So

289
00:22:28,667 --> 00:22:34,088
Speaker 2: sort of you tell it what you wanna do
and it sort of keeps you in line. It's like-

290
00:22:34,128 --> 00:22:34,129
Speaker 1: Mm

291
00:22:34,488 --> 00:22:37,288
Speaker 2: ... "You shouldn't be doing that. Why
are you up at 2:00 A.M. watching anime? You

292
00:22:37,328 --> 00:22:40,628
Speaker 2: shouldn't be doing that. You said you
didn't want to." So it's sort of like

293
00:22:40,688 --> 00:22:45,128
Speaker 2: outsourcing your higher mind to keep
your lower minds in check.

294
00:22:45,808 --> 00:22:46,028
Speaker 1: Mm-hmm.

295
00:22:46,108 --> 00:22:52,088
Speaker 2: And, but ultimately I wanna combine
information from other things. So wearables

296
00:22:52,668 --> 00:22:58,588
Speaker 2: is a big one. Uh, anything that... AI
just wants data. So anytime-- So I have an

297
00:22:58,628 --> 00:23:02,488
Speaker 2: Oura Ring, right? It has all my sleep
data as well as continuous like biometrics.

298
00:23:03,008 --> 00:23:06,948
Speaker 2: Wow, that would be great to
incorporate into the thing that knows everything
about

299
00:23:06,968 --> 00:23:11,108
Speaker 2: my goals and what I'm doing all the
time, and these things just stack on top of each

300
00:23:11,168 --> 00:23:16,748
Speaker 2: other. So I would say anything that
has a consistent high quality amount of data or

301
00:23:16,868 --> 00:23:21,788
Speaker 2: data in large volumes is going to be
very useful.

302
00:23:22,888 --> 00:23:26,948
Speaker 2: If I had to bet on a dark horse
interaction with AI that most people don't see

303
00:23:27,008 --> 00:23:32,788
Speaker 2: coming, I'd say it's probably brain
computer interfaces. Um, most likely in terms of

304
00:23:32,828 --> 00:23:34,848
Speaker 2: consumer EEG.

305
00:23:35,568 --> 00:23:35,788
Speaker 1: Mm-hmm.

306
00:23:35,828 --> 00:23:39,808
Speaker 2: Um, so I think Meta and Apple both
have EEG coming out incorporated into wearables.

307
00:23:40,548 --> 00:23:46,188
Speaker 2: Um, it can do more than you would
expect. Um, so that's probably...

308
00:23:46,548 --> 00:23:51,628
Speaker 2: Plus informa- brain computer
interface information in, you can do that
through

309
00:23:51,688 --> 00:23:55,928
Speaker 2: sensory augmentation. Won't go into
it, but turns out the brain is pretty good at

310
00:23:55,948 --> 00:24:00,628
Speaker 2: getting information into it if you
train it. Um, that one is also pretty

311
00:24:00,668 --> 00:24:05,628
Speaker 2: interesting. So I think people are
gonna start to chafe at how slow information
goes

312
00:24:05,728 --> 00:24:11,268
Speaker 2: out and into their brains and they're
going to be interested in more or less hooking

313
00:24:11,328 --> 00:24:17,236
Speaker 2: up to these systems all the time, and
these are the tools that enable that. You

314
00:24:17,296 --> 00:24:19,936
Speaker 2: may have asked another question, but
those, those are the answers I got for you.

315
00:24:20,676 --> 00:24:25,236
Speaker 1: No, no, that's, that's completely
fine. And, uh, it, it made sense because, you

316
00:24:25,256 --> 00:24:30,136
Speaker 1: know, anything which is, as you said,
generating data, actually these language

317
00:24:30,216 --> 00:24:34,776
Speaker 1: models lives on data, right? So they,
they, they need to suck data to give us back

318
00:24:34,816 --> 00:24:40,616
Speaker 1: what we want. And I think maybe
something related to, to, uh, if we think

319
00:24:40,656 --> 00:24:46,436
Speaker 1: consumer perspective sensors, you
know, anything that comes from cameras and this

320
00:24:46,776 --> 00:24:52,676
Speaker 1: side. In industrial, uh, vertical, I
can think also about sensors and this kind of

321
00:24:52,716 --> 00:24:57,976
Speaker 1: thing. So yeah. Now a, a question
that came to my mind just out of curiosity,

322
00:24:58,156 --> 00:25:03,856
Speaker 1: honestly, like of course I'm too much
into the, uh, you know, anything similar to

323
00:25:03,896 --> 00:25:09,216
Speaker 1: OpenAI and like, and there are a
couple of others, but when it comes to, you
know,

324
00:25:09,316 --> 00:25:15,216
Speaker 1: something like Stable Diffusion,
like, uh, Mindjourney, like, uh, Da- uh, DALL-E
2,

325
00:25:16,796 --> 00:25:22,295
Speaker 1: I want to understand personally. So,
so consider me a, a guy that, that he's not

326
00:25:22,336 --> 00:25:28,156
Speaker 1: from a technical background now. So
for me, what I see is that these are algorithms

327
00:25:28,196 --> 00:25:32,996
Speaker 1: or whatever you want to call them to
generate for you medias, right? So whether it's

328
00:25:33,036 --> 00:25:38,396
Speaker 1: say photos or, you know, kind of
paintings, but

329
00:25:39,236 --> 00:25:45,236
Speaker 1: what are really the real uses or real
world use cases that we can get out

330
00:25:45,296 --> 00:25:50,056
Speaker 1: of, of these like, uh, Stable
Diffusion or, uh, uh,

331
00:25:50,096 --> 00:25:52,196
Speaker 1: Mindjourney?

332
00:25:52,316 --> 00:25:57,576
Speaker 2: I think people would probably be
shocked about what is currently in development.
Um,

333
00:25:57,696 --> 00:26:03,516
Speaker 2: it takes a long time to build
production systems for really-- You get this
Cambrian

334
00:26:03,556 --> 00:26:08,896
Speaker 2: explosion of like people eat up the
easy use cases first. Just slap a user interface

335
00:26:08,936 --> 00:26:12,956
Speaker 2: on ChatGPT or, um, Stable Diffusion.
Cool, there you go. Now you can edit photos.

336
00:26:13,236 --> 00:26:17,256
Speaker 2: That's great. [chuckles] Cool. You
can make people disappear in photos. All right.

337
00:26:17,876 --> 00:26:23,096
Speaker 2: Now what? You can generate art. Who
cares? Um, but you should really consider this

338
00:26:23,256 --> 00:26:23,816
Speaker 2: more like...

339
00:26:25,395 --> 00:26:29,096
Speaker 2: By the way, there are other really
good generative networks. Transformer networks

340
00:26:29,176 --> 00:26:33,096
Speaker 2: are competing against, uh, Diffusion
now, and I think it's only a matter of time

341
00:26:33,136 --> 00:26:37,016
Speaker 2: until they end up working together.
They have different strengths. So generative

342
00:26:37,196 --> 00:26:42,216
Speaker 2: information is really, it's, it's not
slowing down either. I think you should look

343
00:26:42,236 --> 00:26:46,116
Speaker 2: at this not as generating media,
although it does do that, and that will have a

344
00:26:46,136 --> 00:26:51,856
Speaker 2: profound impact, uh, particularly in
terms of needing identity solutions, um,

345
00:26:52,056 --> 00:26:57,196
Speaker 2: and probably the death of the public
anonymous internet in a way. Um, but that's

346
00:26:57,236 --> 00:27:02,236
Speaker 2: okay. Uh, it will bring other good
things too, but it's more like generative

347
00:27:02,276 --> 00:27:08,116
Speaker 2: information. So here's an example
that I, I like to give. You should think of

348
00:27:08,176 --> 00:27:09,716
Speaker 2: these things as being

349
00:27:11,396 --> 00:27:16,756
Speaker 2: really good at generating things to
about 80% or 90% fidelity. Uh, they're not so

350
00:27:16,796 --> 00:27:21,596
Speaker 2: good at generating it to 100%. So
let's say you're, I don't know, you're an
aircraft

351
00:27:21,636 --> 00:27:26,296
Speaker 2: designer at Boeing, or you design
cars. Your, your job is gonna look more like

352
00:27:26,436 --> 00:27:31,856
Speaker 2: sitting back, crossing your arms and
saying, "Okay, I want something that is sleek,

353
00:27:32,056 --> 00:27:36,896
Speaker 2: powerful, uses this last year's
model. Show me 40 different combinations."
Whoop,

354
00:27:37,016 --> 00:27:41,936
Speaker 2: here they are. That one. [chuckles]
Okay. Now given this one, make four different

355
00:27:41,976 --> 00:27:46,936
Speaker 2: variations of this. Run it through
the, you know, the stream or the, um, the tests

356
00:27:47,036 --> 00:27:51,096
Speaker 2: for wind resistance and all this
stuff. Great. I'll come back after lunch. Okay,

357
00:27:51,156 --> 00:27:55,456
Speaker 2: here it is. Here's your model. That
would've taken you how long with clay models or

358
00:27:55,536 --> 00:27:57,716
Speaker 2: even just like going in with computer
models?

359
00:27:58,456 --> 00:27:58,516
Speaker 1: Wow.

360
00:27:58,556 --> 00:28:04,556
Speaker 2: It's un- unbelievable. The amount of
testing an idea, I- the

361
00:28:04,616 --> 00:28:10,496
Speaker 2: i- the cost of testing out ideas is
going to zero, and that's gonna have a very

362
00:28:10,536 --> 00:28:15,476
Speaker 2: big impact on innovation, which is
not really well captured right now or
understood.

363
00:28:16,256 --> 00:28:22,176
Speaker 2: When it costs you $100 or $1,000 or
even $10 to do something and it takes

364
00:28:22,216 --> 00:28:28,016
Speaker 2: you 10, 10 hours, 20 hours to test
it, you don't test that many things.

365
00:28:28,296 --> 00:28:31,816
Speaker 2: You have to go with, by necessity,
the things that look like they're gonna have
ROI.

366
00:28:32,296 --> 00:28:36,536
Speaker 2: But when the cost of testing a
thousand things is basically zero and you can
test a

367
00:28:36,576 --> 00:28:42,216
Speaker 2: thousand things in a minute, things
change dramatically. You, you can explore

368
00:28:42,256 --> 00:28:46,156
Speaker 2: horizontally a lot more than you
could before. You can test out really dumb
ideas,

369
00:28:46,676 --> 00:28:51,396
Speaker 2: and sometimes those dumb ideas turn
out to be really interesting, useful ideas, and

370
00:28:51,456 --> 00:28:55,196
Speaker 2: that's gonna happen everywhere. So
it's not just generating media, it's generating

371
00:28:55,816 --> 00:29:01,336
Speaker 2: information of any kind. Uh, but it
will be-- The reason we see media is because we

372
00:29:01,376 --> 00:29:06,076
Speaker 2: have a public open internet and
everyone dumps all their data online, so guess
what

373
00:29:06,096 --> 00:29:10,616
Speaker 2: they trained on, is the stuff that
there was a lot of data for. But there is data

374
00:29:10,696 --> 00:29:15,396
Speaker 2: for other stuff. It's a little harder
to get and it's a little less lucrative

375
00:29:15,476 --> 00:29:21,096
Speaker 2: upfront and a little harder to build,
but it's coming. So, uh, any

376
00:29:21,156 --> 00:29:26,796
Speaker 2: information fundamentally, whether
it's science experiments, you know, generating

377
00:29:26,836 --> 00:29:31,456
Speaker 2: novel drug compounds, all of this
stuff is just information. Genetics.

378
00:29:32,976 --> 00:29:37,416
Speaker 2: So if you want to test something out
to just 80% or 90% fidelity really, really,

379
00:29:37,516 --> 00:29:42,416
Speaker 2: really fast, and then that last 10%,
20% is still done with humans and maybe regular

380
00:29:42,476 --> 00:29:47,616
Speaker 2: computers, that's a big change. It's
a couple orders of magnitude, I think, in terms

381
00:29:47,696 --> 00:29:51,036
Speaker 2: of like speed of innovation. So
that's, that's a huge use case for the
generative

382
00:29:51,096 --> 00:29:51,536
Speaker 2: networks.

383
00:29:53,196 --> 00:29:58,316
Speaker 1: That's, uh, you know, like it, it's,
it's like very enlightening, I would say,

384
00:29:58,736 --> 00:30:03,236
Speaker 1: thinking about doing simulations
using these technologies because they can
generate

385
00:30:03,336 --> 00:30:09,156
Speaker 1: different, let's say, prototype for
you. But there are some people that

386
00:30:09,196 --> 00:30:15,056
Speaker 1: they are arguing that, um, we might
be too

387
00:30:15,096 --> 00:30:20,560
Speaker 1: much depending on the existing
knowledge that already these- ... models they
have,

388
00:30:21,060 --> 00:30:26,220
Speaker 1: so it will be lazy to generate new
content for the language models to use. What do

389
00:30:26,260 --> 00:30:27,160
Speaker 1: you think about that?

390
00:30:29,580 --> 00:30:29,980
Speaker 2: I think

391
00:30:31,860 --> 00:30:37,840
Speaker 2: how much does it cost to get the best
aircraft modeler or

392
00:30:37,900 --> 00:30:43,520
Speaker 2: the best car modeler to do something?
A lot of money. How much does it cost to get

393
00:30:43,680 --> 00:30:49,660
Speaker 2: 80 to 90% [chuckles] of the best?
Still probably quite a bit. So the difference
is,

394
00:30:50,160 --> 00:30:55,600
Speaker 2: okay, yes, if even if we-- let's say
I'll, I'll buy that argument, and we can only

395
00:30:55,640 --> 00:31:01,360
Speaker 2: get to even 80% of the average of any
skill, given it has enough training

396
00:31:01,420 --> 00:31:05,900
Speaker 2: data. Who cares, right? All it's
doing is repeating existing patterns. Well, what

397
00:31:05,960 --> 00:31:11,560
Speaker 2: matters is that I don't have to learn
a skill. I can have 0% car modeling knowledge,

398
00:31:11,660 --> 00:31:16,300
Speaker 2: and now I can model a car. That's a
very big deal. Uh, just in the practical

399
00:31:16,400 --> 00:31:21,380
Speaker 2: perspective, like for my stories,
right? I can do the cover art. I can't do cover

400
00:31:21,440 --> 00:31:27,220
Speaker 2: art. Before this, I spent $250 or
$500 for like incredible concept art, and it

401
00:31:27,260 --> 00:31:32,140
Speaker 2: took weeks of back and forth, and it
was a huge pain, and half the time it doesn't

402
00:31:32,160 --> 00:31:36,140
Speaker 2: end up looking right. Now I can just,
just, I don't know, I'll just throw stuff at

403
00:31:36,180 --> 00:31:39,360
Speaker 2: the wall. I'll just do it for four
hours, and I've generated a couple hundred

404
00:31:39,440 --> 00:31:44,960
Speaker 2: pictures, and it costs me basically
nothing. You can imagine that for any field.

405
00:31:45,020 --> 00:31:50,340
Speaker 2: Like, you don't have to learn a skill
to do it at 80% capacity of

406
00:31:50,800 --> 00:31:51,240
Speaker 2: average-

407
00:31:51,520 --> 00:31:51,521
Speaker 1: Mm-hmm

408
00:31:51,520 --> 00:31:56,660
Speaker 2: ... like the average professional.
That's a very big deal. And so you don't even

409
00:31:56,720 --> 00:32:01,060
Speaker 2: have to advance state-of-the-art in
order for it to matter significantly. It's about

410
00:32:01,080 --> 00:32:02,740
Speaker 2: democratization of skill set.

411
00:32:05,560 --> 00:32:10,860
Speaker 1: Well, that, that's another point of
view, I would say, which is also valid. Uh,

412
00:32:11,060 --> 00:32:14,960
Speaker 1: Stephen, can you tell me, because I
know you, you write sci-fi also as well, right?

413
00:32:15,060 --> 00:32:20,580
Speaker 1: So, uh, can you tell us a little bit
more about this side? Uh, I, I'm sure that I

414
00:32:20,620 --> 00:32:24,280
Speaker 1: have a lot of my audience who would
be interested to, to know more about it.

415
00:32:24,600 --> 00:32:30,380
Speaker 2: Sure. Yeah, it's just for fun. Um,
but I do occasionally try to write--

416
00:32:30,560 --> 00:32:34,460
Speaker 2: Sometimes there's a combination of
these. So it's sci-fi and fantasy, so the
fantasy

417
00:32:34,480 --> 00:32:40,000
Speaker 2: is mostly for fun. I always ha- try
to have a lesson. Um, so I've written a story

418
00:32:40,080 --> 00:32:46,060
Speaker 2: recently called "Reverie." That is a
story about I think that humanity doesn't have

419
00:32:46,120 --> 00:32:50,880
Speaker 2: a lot of great things to run towards,
like collective goals. I think we have a lot

420
00:32:50,900 --> 00:32:55,020
Speaker 2: of things we're trying to avoid, but
that's not super motivating. Who gets up every

421
00:32:55,120 --> 00:32:58,880
Speaker 2: day saying, "Oh, I hope I don't--
Hope I'm not in pain today"? That's not super

422
00:32:58,980 --> 00:33:04,080
Speaker 2: motivating. Um, so I, I tried to come
up with here's where I think we should try to

423
00:33:04,120 --> 00:33:09,720
Speaker 2: go as a species. Um, I'm writing a
story now called "Hellmakers." This is

424
00:33:09,780 --> 00:33:14,800
Speaker 2: about... This is why I smiled when
you mentioned, um, s- I think you mentioned

425
00:33:14,860 --> 00:33:18,140
Speaker 2: something about, um, bots, automated
bots.

426
00:33:18,920 --> 00:33:18,921
Speaker 1: Yeah.

427
00:33:18,980 --> 00:33:24,280
Speaker 2: So already I've seen people talk
about somebody has already had the bright idea.
You

428
00:33:24,340 --> 00:33:28,220
Speaker 2: know what would be great? We could--
We need to open source these models because,

429
00:33:28,460 --> 00:33:31,460
Speaker 2: you know, people probably don't want
us to have access to this. It sh- everything

430
00:33:31,500 --> 00:33:35,680
Speaker 2: should be free and open. You know
what else we should do? We should put this on

431
00:33:35,720 --> 00:33:40,220
Speaker 2: decentralized computing platforms.
This is a good idea. And I'm just like, this is

432
00:33:40,260 --> 00:33:41,400
Speaker 2: the worst possible idea.

433
00:33:41,720 --> 00:33:41,721
Speaker 1: Wow.

434
00:33:41,760 --> 00:33:46,700
Speaker 2: Let me write a story to tell you why.
So it's a, it's basically a story about the

435
00:33:46,760 --> 00:33:52,280
Speaker 2: founder of a decentralized digital
autonomous worker protocol who they just start

436
00:33:52,320 --> 00:33:58,260
Speaker 2: like a decentralized computing, uh,
blockchain protocol, right? Um, the goal

437
00:33:58,340 --> 00:34:02,440
Speaker 2: is that it uses a consensus
mechanism, so you can't turn it off easily
unless you

438
00:34:02,480 --> 00:34:04,320
Speaker 2: can turn off all the nodes on the
network.

439
00:34:05,000 --> 00:34:05,060
Speaker 1: Oh.

440
00:34:05,600 --> 00:34:10,240
Speaker 2: You can't turn it off. And it's
permissionless, so anyone can insert money, and
you

441
00:34:10,300 --> 00:34:13,940
Speaker 2: insert a prompt, pull a model off the
shelf, and you go tell it to do something, and

442
00:34:14,000 --> 00:34:19,400
Speaker 2: it will go do whatever until it runs
out of money. Uh, and you can't turn it off. So

443
00:34:19,580 --> 00:34:25,560
Speaker 2: I show this is probably not a good
idea. So the punchline of the story is that

444
00:34:25,620 --> 00:34:30,280
Speaker 2: the, the, uh, creator of the protocol
ends up using this to get back at their

445
00:34:30,340 --> 00:34:35,960
Speaker 2: co-founders who cuts them out of a
deal and, as the name implies, creates a

446
00:34:36,340 --> 00:34:41,740
Speaker 2: bot that literally just publicly
tries to ruin this other person in

447
00:34:41,780 --> 00:34:46,740
Speaker 2: perpetuity and cannot be turned off.
So this is not good. We should think very

448
00:34:46,800 --> 00:34:50,960
Speaker 2: carefully about not having an off
switch for our machines. It just seemed like a

449
00:34:51,020 --> 00:34:56,080
Speaker 2: very dumb idea to put them on a box
that you can't turn off frankly.

450
00:34:56,180 --> 00:35:01,920
Speaker 1: [clears throat] Um, a question that I
ask now, you-- it, it became

451
00:35:02,000 --> 00:35:07,860
Speaker 1: a, um, traditional question for me.
Like, um, do you think that

452
00:35:08,180 --> 00:35:13,740
Speaker 1: AI is allowing all of us, humanity,
to reach, to reach

453
00:35:13,900 --> 00:35:14,600
Speaker 1: singularity?

454
00:35:20,500 --> 00:35:22,280
Speaker 2: Maybe. Um-

455
00:35:22,660 --> 00:35:26,160
Speaker 1: That's a, that's a philosophical
question, I know, so [chuckles]

456
00:35:26,240 --> 00:35:31,460
Speaker 2: The key is in what you said at the
end, which is all of us and humanity to reach
the

457
00:35:31,500 --> 00:35:36,100
Speaker 2: singularity. Yes. I actually think
that positive outcomes are the most likely

458
00:35:36,200 --> 00:35:41,160
Speaker 2: scenario in the long term, and
there's a lot we can do about it, and a lot of
things

459
00:35:41,180 --> 00:35:44,620
Speaker 2: that we could be doing about it that
are actually quite feasible that are probably

460
00:35:44,660 --> 00:35:50,560
Speaker 2: not well known, so it's something I'm
focused on. But yes, I think technically

461
00:35:50,640 --> 00:35:56,040
Speaker 2: speaking, yes, this is the way you
would need scaled intelligence. And I think the

462
00:35:56,080 --> 00:36:00,800
Speaker 2: most important part of AI as moving
us towards singularity is that

463
00:36:02,060 --> 00:36:06,200
Speaker 2: maybe is also underappreciated, is
that it doesn't have cognitive biases. Uh,
unlike

464
00:36:06,220 --> 00:36:11,680
Speaker 2: our own brains, we can reprogram it,
and we can program away the blind spots that we

465
00:36:11,740 --> 00:36:16,440
Speaker 2: know we have. It doesn't matter if
you know about anchoring bias or recency bias,

466
00:36:17,260 --> 00:36:21,900
Speaker 2: um, if you still d- you experience
them whether you want them or not. You still
make

467
00:36:21,940 --> 00:36:26,712
Speaker 2: dumb choices all the-- If you're
tired- And hungry, you're gonna make bad
choices.

468
00:36:26,832 --> 00:36:31,492
Speaker 2: Like, this is just how the human
brain works. You can't program around it. But

469
00:36:31,512 --> 00:36:35,332
Speaker 2: that's not the same for our machines,
which means that we should be able to program

470
00:36:35,372 --> 00:36:40,712
Speaker 2: them to our ideals. And as long as we
can keep advancing them, I don't see why, you

471
00:36:40,752 --> 00:36:45,452
Speaker 2: know, you can't have wonderful,
incredible outcomes from that. So in short, yes,
I--

472
00:36:45,612 --> 00:36:47,972
Speaker 2: but I think it's neutral. It depends
on what we do.

473
00:36:49,072 --> 00:36:54,812
Speaker 1: That's, that's fair enough, I would
say. Um, be- before I will, you know, keep it

474
00:36:54,852 --> 00:36:59,332
Speaker 1: for you to do a final conclusion, not
related to AI, related because you are a

475
00:36:59,372 --> 00:37:03,392
Speaker 1: solopreneur, right? So how, how,

476
00:37:05,072 --> 00:37:08,692
Speaker 1: how do you, do you see it? Like,
what's your experience being a solopreneur?
Because

477
00:37:08,772 --> 00:37:13,392
Speaker 1: again, this is the, this is the
second, uh, common question I'm asking to my
guests

478
00:37:13,452 --> 00:37:18,552
Speaker 1: if they are solopreneurs, like why
you opted to be a solopreneur, like why you're

479
00:37:18,612 --> 00:37:23,072
Speaker 1: not part of-- I, I know that you, you
worked for some companies before, and I did by

480
00:37:23,112 --> 00:37:29,031
Speaker 1: the way as well. I'm a solopreneur
kind of now. So tell me your experience as a

481
00:37:29,072 --> 00:37:33,212
Speaker 1: solopreneur and why we are seeing
more solopreneurs, uh, in, in the field.

482
00:37:36,952 --> 00:37:40,392
Speaker 2: So I'll answer the last one first
'cause it's, it's fairly simple. I think we're

483
00:37:40,432 --> 00:37:44,252
Speaker 2: seeing more-- people are more
isolated in general, but I think we're also more

484
00:37:44,292 --> 00:37:48,392
Speaker 2: empowered by technology in general.
Like I mentioned, like I, I can do a lot of

485
00:37:48,432 --> 00:37:53,232
Speaker 2: things I don't know how to do thanks
to advances in, in AI and other software,

486
00:37:53,312 --> 00:37:53,412
Speaker 2: right?

487
00:37:54,192 --> 00:37:54,272
Speaker 1: Yeah.

488
00:37:54,292 --> 00:37:58,812
Speaker 2: Um, I don't know how to compute
compound interest. I mean, I could do it by hand
if

489
00:37:58,872 --> 00:38:03,632
Speaker 2: you gave me the formula and a lot of
time, but a spreadsheet can do it. Uh, so I'm

490
00:38:03,652 --> 00:38:08,412
Speaker 2: enabled to do a lot by myself, and
because I'm working in software and AI, which is

491
00:38:08,452 --> 00:38:14,292
Speaker 2: like software for software, it's-- I
can-- I have a lot of leverage. Um,

492
00:38:14,772 --> 00:38:20,552
Speaker 2: so I am enabled to do it. Um, but
also at the same time, it's often

493
00:38:20,612 --> 00:38:24,292
Speaker 2: easier just to pay contractors, and I
find that there is a sort of cultural

494
00:38:24,352 --> 00:38:26,032
Speaker 2: zeitgeist movement towards,

495
00:38:28,012 --> 00:38:32,312
Speaker 2: uh, people who want their own
autonomy, and they wanna interface with other
people

496
00:38:32,372 --> 00:38:37,832
Speaker 2: as people, and they don't want to be
an employee. So I found it very easy to just,

497
00:38:37,932 --> 00:38:41,012
Speaker 2: "Hey, what's your rate? Like, let's
figure out a deal, let's figure out a

498
00:38:41,052 --> 00:38:45,852
Speaker 2: partnership." And that has been the
easier way to go. So that's the answer to the

499
00:38:45,872 --> 00:38:50,552
Speaker 2: second part. So the first part, my
experience as a solopreneur, I think that

500
00:38:51,752 --> 00:38:56,232
Speaker 2: the challenge, especially with AI,
right? So selling software development is fairly

501
00:38:56,312 --> 00:39:01,652
Speaker 2: straightforward because people
understand kind of how to, how to value it.

502
00:39:02,272 --> 00:39:07,992
Speaker 2: The biggest challenge of getting paid
to build stuff with AI is how to help

503
00:39:08,032 --> 00:39:13,172
Speaker 2: people understand the opportunities
that they have and how to help them value it.

504
00:39:13,872 --> 00:39:18,112
Speaker 2: It's-- There's a lot of work that has
to be done. You don't have to do a lot of work

505
00:39:18,152 --> 00:39:22,592
Speaker 2: to tell people about regular software
systems. They already know how to reason about

506
00:39:22,632 --> 00:39:26,912
Speaker 2: it. There's a lot of competitors they
know what to compare it to. It's very simple.

507
00:39:27,492 --> 00:39:33,352
Speaker 2: But for, for AI, there's a long
conversation of education. Okay, here's what is

508
00:39:33,372 --> 00:39:38,552
Speaker 2: possible. Okay, now I need to fully
understand your business so that I can analyze,

509
00:39:39,232 --> 00:39:43,732
Speaker 2: okay, where's the opportunity? And
then from there, I have to help you understand

510
00:39:43,772 --> 00:39:47,032
Speaker 2: what the value is, that we can
justify the cost, which is higher than regular

511
00:39:47,112 --> 00:39:49,752
Speaker 2: software. There's a lot of steps. So-

512
00:39:49,832 --> 00:39:49,932
Speaker 1: Yeah

513
00:39:49,972 --> 00:39:53,712
Speaker 2: ... it's a very high trust thing.
Also, I noticed that

514
00:39:55,272 --> 00:39:59,552
Speaker 2: most businesses-- So again, this is
just money from consulting, not building my own

515
00:39:59,572 --> 00:40:04,092
Speaker 2: products. Most businesses would
benefit... They're not even fully utilizing

516
00:40:04,172 --> 00:40:09,632
Speaker 2: software. They're n- definitely not
even fully using non-software solutions. And so

517
00:40:09,712 --> 00:40:14,532
Speaker 2: AI is often like, why would you bring
this really complex solution in? 'Cause it's

518
00:40:14,572 --> 00:40:19,392
Speaker 2: sexy, it's got buzzwords. But I've
found so far, for the most part,

519
00:40:20,732 --> 00:40:26,252
Speaker 2: price optimization, retention
prediction, uh, collaborative filtering, just--

520
00:40:27,012 --> 00:40:32,432
Speaker 2: even just regression, like linear s-
um, logistic regression, just statistical

521
00:40:32,512 --> 00:40:37,592
Speaker 2: models. These are often like 80/20,
so most people don't need the big guns.

522
00:40:38,872 --> 00:40:43,872
Speaker 2: That is the cool stuff for sure. I'd
say generative AI or the large language models

523
00:40:43,911 --> 00:40:49,412
Speaker 2: changes things a little bit. But by
and large, you don't need AI for most solutions.

524
00:40:50,152 --> 00:40:54,812
Speaker 2: So a lot of the challenge is finding
the people who really do have a slam dunk use

525
00:40:54,852 --> 00:41:00,812
Speaker 2: case and building the trust, telling
the story. So yeah,

526
00:41:01,292 --> 00:41:05,012
Speaker 2: that's, that's what it's like, is
that there's a lot of legwork. That's sort of
the

527
00:41:05,112 --> 00:41:05,852
Speaker 2: AI side.

528
00:41:07,932 --> 00:41:11,972
Speaker 1: Yeah. That's fair enough, I would
say, and, uh, I have to agree with a lot of the

529
00:41:12,072 --> 00:41:16,812
Speaker 1: points that you mentioned, Steven.
Steven, any final thing you-- maybe something I

530
00:41:16,852 --> 00:41:21,112
Speaker 1: didn't ask you, you want to, to, to
say or share before we, we close?

531
00:41:23,212 --> 00:41:29,092
Speaker 2: Um, well, my producer and, um, the
social media growth manager would

532
00:41:29,292 --> 00:41:33,012
Speaker 2: probably hit me over the head if I
didn't say go to the YouTube channel.

533
00:41:33,732 --> 00:41:33,972
Speaker 1: Yes, please.

534
00:41:34,032 --> 00:41:35,992
Speaker 2: Seeking Minima. [chuckles]

535
00:41:36,052 --> 00:41:36,452
Speaker 1: Yes.

536
00:41:36,672 --> 00:41:41,992
Speaker 2: Um, but-- So I mean, do that if you
want to hear more words like this.

537
00:41:43,032 --> 00:41:43,033
Speaker 2: But-

538
00:41:43,092 --> 00:41:46,832
Speaker 1: Actually, I wanted to ask where
people can find more about you. [laughs]

539
00:41:46,892 --> 00:41:47,012
Speaker 2: Ah, yeah.

540
00:41:47,032 --> 00:41:48,112
Speaker 1: Like this one. Yeah.

541
00:41:48,432 --> 00:41:53,912
Speaker 2: I, I, I'll get better at this, guys.
Um, but so what I wanted to say is, uh,

542
00:41:54,492 --> 00:41:58,652
Speaker 2: I wanna-- the whole purpose of
Seeking Minima in general is this is not a
for-profit

543
00:41:58,732 --> 00:42:02,812
Speaker 2: thing. Like, I can just build
software. Like, there's no point recording
videos or

544
00:42:03,172 --> 00:42:08,312
Speaker 2: talking to people like you for, um,
for a living. That's not what I wanna do. I am

545
00:42:08,672 --> 00:42:14,432
Speaker 2: trying to help people use technology
and to create better outcomes for everybody,

546
00:42:15,212 --> 00:42:20,392
Speaker 2: and I think that we are at a fairly
pivotable, pivotal moment, and I see a lot of,

547
00:42:20,892 --> 00:42:21,852
Speaker 2: despite some of the

548
00:42:23,392 --> 00:42:28,352
Speaker 2: pessimistic takes so far, that's not
really how I look at the world. I think that we

549
00:42:28,372 --> 00:42:33,760
Speaker 2: have a lot of opportunity- To make a
big difference right now. And I want to help

550
00:42:33,800 --> 00:42:38,620
Speaker 2: people realize that they don't need
to stand here and look scared. Um, there's a lot

551
00:42:38,660 --> 00:42:44,580
Speaker 2: we can do in terms of, uh, AI
alignment, um, wealth inequality. We

552
00:42:44,620 --> 00:42:48,520
Speaker 2: have a lot of tools at our disposal,
and we're at the really sweet spot where

553
00:42:48,560 --> 00:42:52,520
Speaker 2: there's... Most of this stuff is
greenfield. It's brand new. We have a lot of

554
00:42:52,580 --> 00:42:57,180
Speaker 2: opportunity to use it well, to put
common sense rules in place, and to really set
it

555
00:42:57,260 --> 00:43:02,359
Speaker 2: up to work for us instead of against
us or for a select few peopl- few people.

556
00:43:03,360 --> 00:43:07,400
Speaker 2: And I think that, yeah, I'm gonna
continue exploring those things. I have a lot of

557
00:43:07,440 --> 00:43:12,120
Speaker 2: concrete ideas, but also I just like
to synthesize other people's ideas. There's no

558
00:43:12,180 --> 00:43:17,280
Speaker 2: need to be terrified of the future
because, uh, we have a lot of control over it,

559
00:43:17,800 --> 00:43:20,640
Speaker 2: and we're the ones who control it. I
don't know. People seem to forget that. As if

560
00:43:20,660 --> 00:43:26,200
Speaker 2: by the way, the future, what future
happens is based on what we do. So

561
00:43:26,660 --> 00:43:30,700
Speaker 2: if you don't like it, you should do
something about it, and that's, that's what I'm

562
00:43:30,740 --> 00:43:35,460
Speaker 2: doing. If you're interested in that
too, then yeah, I'd love to keep talking with

563
00:43:35,500 --> 00:43:35,580
Speaker 2: you.

564
00:43:36,720 --> 00:43:41,340
Speaker 1: That's really great, and I think we
are on a same mission, I would say, Steven.

565
00:43:41,440 --> 00:43:46,480
Speaker 1: Because one of the reasons I'm doing
all, you know, this podcast, I'm, you know,

566
00:43:47,600 --> 00:43:53,020
Speaker 1: creating a lot of content is to raise
awareness. This is first. And tell people, you

567
00:43:53,060 --> 00:43:58,000
Speaker 1: know, like, hey, uh, you know, it's
not like only the scary stuff you see in the

568
00:43:58,080 --> 00:44:02,520
Speaker 1: media. You know, like people gonna
lose their jobs. I don't know what. You know,
all

569
00:44:02,580 --> 00:44:08,480
Speaker 1: this negativity. So I'm trying to put
some light on the positive side of the

570
00:44:08,540 --> 00:44:12,020
Speaker 1: technology, and not just AI by the
way, any technology. Uh-

571
00:44:12,220 --> 00:44:12,280
Speaker 2: Sure

572
00:44:12,300 --> 00:44:17,000
Speaker 1: ... 'cause I, I, I love technology
myself, like since I was a child. And I always

573
00:44:17,040 --> 00:44:23,020
Speaker 1: believe that technology's main goal
is to take us forward, make

574
00:44:23,060 --> 00:44:28,300
Speaker 1: our lives easy. From business
perspective, it's always, I'm repeating myself
on

575
00:44:28,360 --> 00:44:34,160
Speaker 1: multiple episodes, increase revenues,
increase customer base, decrease

576
00:44:34,260 --> 00:44:39,440
Speaker 1: churn, all these nice things that,
uh, business people like to, to, to, to hear.
And

577
00:44:39,780 --> 00:44:45,340
Speaker 1: AI is no different than this. So AI
will make, you know, um, your business thrive

578
00:44:45,440 --> 00:44:49,220
Speaker 1: and, uh, better. And for individuals,
I'm saying exactly what you mentioned at the

579
00:44:49,260 --> 00:44:54,240
Speaker 1: beginning also as well. Guys, like if
you are scared of something, go study it. Like

580
00:44:54,280 --> 00:44:58,260
Speaker 1: this is the best thing you can do.
Because if you just keep watching it and do

581
00:44:58,300 --> 00:45:02,720
Speaker 1: nothing, yeah, you're gonna be
scared. I, sorry, I cannot help. This is why I'm

582
00:45:02,760 --> 00:45:08,100
Speaker 1: trying to help you in generating this
content as much as I can, of course. Um,

583
00:45:08,660 --> 00:45:12,860
Speaker 1: so yeah, this is why we are aligned,
and this is actually why I wanted to interview

584
00:45:12,900 --> 00:45:17,780
Speaker 1: you, Steven, because when I read, uh,
your bio, this what attracted me. And by the

585
00:45:17,840 --> 00:45:22,220
Speaker 1: way, this is I do with all my guests.
I will share, you know, the YouTube channel,

586
00:45:22,280 --> 00:45:26,040
Speaker 1: like anything else I can find if you
want to share that with me also as well. So

587
00:45:26,080 --> 00:45:30,740
Speaker 1: that would go into the description of
the episode and desc- the description on the

588
00:45:30,800 --> 00:45:35,900
Speaker 1: YouTube also as well. Well, we, we
came to an end, Steven. Thank you very much for

589
00:45:35,960 --> 00:45:39,380
Speaker 1: being my guest today. I really
appreciate the time. And-

590
00:45:39,420 --> 00:45:39,860
Speaker 2: My pleasure

591
00:45:39,900 --> 00:45:44,140
Speaker 1: ... hope... Thank you very much. And
for my audience, whether you are watching this

592
00:45:44,220 --> 00:45:49,860
Speaker 1: on YouTube or you are listening using
your favorite podcasting platform,

593
00:45:50,340 --> 00:45:56,200
Speaker 1: don't forget to subscribe. And as
usual, I'm telling you guys, I would love

594
00:45:56,300 --> 00:46:00,700
Speaker 1: to hear your feedback about this
episode or about the show in general. If you are

595
00:46:00,740 --> 00:46:06,240
Speaker 1: interested to be a guest like Steven
today, like don't be shy. Please come up with

596
00:46:06,580 --> 00:46:10,340
Speaker 1: the idea that you want to discuss,
and I will be more than happy to have you as

597
00:46:10,380 --> 00:46:15,780
Speaker 1: guest with me. And as usual, until we
meet next time, thank you very much, and we

598
00:46:15,820 --> 00:46:16,300
Speaker 1: will see you soon.

599
00:46:16,300 --> 00:46:25,860
Speaker 1: [outro music]