1
00:00:00,160 --> 00:00:03,919
Today on the data driven podcast, we have the privilege of hosting

2
00:00:03,919 --> 00:00:07,460
none other than Jeremy Utley. Now, Jeremy

3
00:00:07,520 --> 00:00:11,205
isn't just any guest. He's an academic marvel

4
00:00:11,205 --> 00:00:14,885
and entrepreneurial spirit rolled into 1. Hailing

5
00:00:14,885 --> 00:00:18,680
from the prestigious corridors of Stanford as an adjunct professor, he's

6
00:00:18,680 --> 00:00:22,120
the kind of chap who educates the future disruptors of Silicon

7
00:00:22,120 --> 00:00:25,800
Valley. He is here to tell us how to get the most out of

8
00:00:25,800 --> 00:00:28,955
generative AI. Now on to the show.

9
00:00:31,654 --> 00:00:35,335
Hello, and welcome back to Data Driven, the podcast where we explore the emergent fields

10
00:00:35,335 --> 00:00:38,940
of artificial intelligence, data engineering, and data science,

11
00:00:39,640 --> 00:00:43,480
and all the associated technologies. With me today is

12
00:00:43,480 --> 00:00:46,985
Jeremy Utley, who is a, adjunct

13
00:00:46,985 --> 00:00:50,685
professor, venture investor, and co author of the book,

14
00:00:51,225 --> 00:00:55,030
Idea Flow, The Only Business Metric That Matters. Welcome to the

15
00:00:55,030 --> 00:00:58,730
show, Jeremy. Thanks for having me. Hey, no problem.

16
00:00:59,670 --> 00:01:03,245
So Stanford. That's kind of a big

17
00:01:03,245 --> 00:01:07,085
deal. It's a it's a special place. Yeah. I'm

18
00:01:07,085 --> 00:01:10,225
just trying to not get found out. I'm sure it's, you know, similar to,

19
00:01:11,259 --> 00:01:15,020
the, the guy on Office Space. Right? At some point, there'll the clerical error

20
00:01:15,020 --> 00:01:18,700
will be revealed. You'll you'll know when they move you to the

21
00:01:18,700 --> 00:01:22,415
basement. Right? Exactly. Exactly. Yeah. But I've been teaching at

22
00:01:22,415 --> 00:01:26,255
Stanford since 2009, and I've been delighted to get to learn alongside some of

23
00:01:26,255 --> 00:01:29,909
the most incredible students in the world and and get to study some of those

24
00:01:29,909 --> 00:01:33,590
incredible innovators in the world. So, not just I may be a

25
00:01:33,590 --> 00:01:36,950
professor or an adjunct professor, but I really consider myself to be a front row

26
00:01:36,950 --> 00:01:40,524
student in in the classroom alongside my students. Very

27
00:01:40,524 --> 00:01:43,505
cool. Very cool. So, what

28
00:01:44,420 --> 00:01:47,320
what is the most important metric? I'll start right there.

29
00:01:49,140 --> 00:01:52,775
Well, the most important metric we call idea

30
00:01:52,775 --> 00:01:56,215
flows, the only business metric that matters. And the reason that we make that bold

31
00:01:56,215 --> 00:01:59,895
claim is because it's the only measure of your

32
00:01:59,895 --> 00:02:03,479
team's capacity to solve problems. And the the only

33
00:02:03,479 --> 00:02:07,240
constant in our day to day lives is problems. In in our businesses, I

34
00:02:07,240 --> 00:02:10,974
don't know, a single business that is facing a day without problems. And

35
00:02:10,974 --> 00:02:14,495
so if you think about problems as the constant, then your

36
00:02:14,495 --> 00:02:18,310
team's capacity to solve problems is really the most important thing

37
00:02:18,310 --> 00:02:21,990
that you should be measuring. And yet, nobody really even knows how to

38
00:02:21,990 --> 00:02:25,430
measure it. And so we talk about idea flow as the as the way to

39
00:02:25,430 --> 00:02:27,850
measure a team's capacity to solve problems.

40
00:02:29,584 --> 00:02:33,185
Interesting. Interesting. And is this is this changing now when we have the

41
00:02:33,185 --> 00:02:36,325
reality of AI assisted teams?

42
00:02:36,980 --> 00:02:40,740
Yeah. Yeah. Absolutely. That's it's a really insightful question. Yes.

43
00:02:40,740 --> 00:02:44,100
It does change or sorry. It has the

44
00:02:44,100 --> 00:02:47,625
potential to change. And yet, what our research suggests that we've

45
00:02:47,625 --> 00:02:51,385
conducted over the last year or so, is that sadly, it

46
00:02:51,385 --> 00:02:55,069
actually doesn't change in practice. In theory, it could change but in

47
00:02:55,069 --> 00:02:58,770
practice it often doesn't. What's interesting, so what

48
00:02:59,150 --> 00:03:02,855
what are the barriers to this? Right? Because I have some thoughts on this. I

49
00:03:02,855 --> 00:03:06,155
know that a number of companies have basically outright

50
00:03:06,455 --> 00:03:09,460
banned, use of AI tools

51
00:03:10,260 --> 00:03:14,020
with good intentions, right, because the privacy policies, etcetera, etcetera,

52
00:03:14,020 --> 00:03:17,400
but in reality, people are copying and pasting sensitive stuff anyway.

53
00:03:17,780 --> 00:03:21,445
So, it seems like banning something outright

54
00:03:21,445 --> 00:03:25,125
doesn't always work in a number of areas. But

55
00:03:25,125 --> 00:03:28,900
what what are the barriers? Right? Because it it it can, like

56
00:03:28,900 --> 00:03:32,040
you said, but in practice, what's what's what are the blockers?

57
00:03:33,140 --> 00:03:36,520
Ultimately, it's it's human psychology, really, is what what's,

58
00:03:36,995 --> 00:03:40,295
the challenge. It turns out that our expectations of the technology,

59
00:03:41,395 --> 00:03:45,159
are hamstringing our ability to make use of it. Because

60
00:03:45,379 --> 00:03:48,599
we're approaching the technology. Most teams that we studied

61
00:03:48,980 --> 00:03:52,745
approach the technology as an oracle. It's almost like a search

62
00:03:52,745 --> 00:03:55,885
box. It's gonna give them the best answer. Right?

63
00:03:56,505 --> 00:04:00,025
And that's the wrong way to approach the technology. It does feel somewhat

64
00:04:00,025 --> 00:04:03,739
magical when you type in, you know, an enigmatic query

65
00:04:03,800 --> 00:04:07,319
and get a seemingly intelligent response. I mean, that feels

66
00:04:07,319 --> 00:04:11,155
magical, but the teams that do that underperform. The teams

67
00:04:11,155 --> 00:04:13,655
that overperform are the teams that

68
00:04:14,674 --> 00:04:18,274
treat generative AI not as an oracle, but as a thought

69
00:04:18,274 --> 00:04:21,820
partner, as as a conversation partner, and iteratively

70
00:04:22,120 --> 00:04:25,740
work together with the AI to discover a better answer.

71
00:04:25,995 --> 00:04:29,755
And the irony of that is it's not very magical, actually. It feels

72
00:04:29,755 --> 00:04:33,435
like work. And yet, where teams that

73
00:04:33,435 --> 00:04:36,620
treat AI as a conversation partner arrive is

74
00:04:37,080 --> 00:04:40,700
light years better than teams that treat

75
00:04:41,000 --> 00:04:43,420
AI as an oracle perform.

76
00:04:44,675 --> 00:04:48,355
Interesting. So what are the what I think I know what you're

77
00:04:48,355 --> 00:04:51,635
getting at in terms of treating it like an oracle versus treating it like a

78
00:04:51,635 --> 00:04:55,340
conversation. Because I've seen that as I do more and more of this, I

79
00:04:55,340 --> 00:04:59,100
hate the term prompt engineering. I hate the strong word. I have mixed feelings about

80
00:04:59,100 --> 00:05:02,915
the term prompt engineering because there is no one single prompt to rule them

81
00:05:02,915 --> 00:05:06,515
all. Mhmm. At least that's been my experience where you kind of you kind

82
00:05:06,515 --> 00:05:10,170
of it's like a conversation, like, you're having. It's not a

83
00:05:10,170 --> 00:05:12,650
person I know. It's not a person I know. It's not a But it's but

84
00:05:12,650 --> 00:05:16,410
but it is a mindset, Frank. That's the thing. It's a mindset. And people

85
00:05:16,410 --> 00:05:20,185
don't come with the mindset of I want to have a conversation. People are lazy.

86
00:05:20,185 --> 00:05:23,485
Right? So so Herbert Simon, back in 1954,

87
00:05:23,945 --> 00:05:27,389
won the Nobel prize for what he deemed satisficing, which

88
00:05:27,389 --> 00:05:31,230
was the human tendency to settle for good enough. Right? And in most

89
00:05:31,230 --> 00:05:33,669
of our lives, it's fine. I need a good enough pair of jeans, I need

90
00:05:33,669 --> 00:05:36,835
a good enough cup of coffee, whatever it is. Right? But when we're trying to

91
00:05:36,835 --> 00:05:40,555
solve problems, good enough sometimes is is okay. But often, especially

92
00:05:40,555 --> 00:05:43,835
when it pertains to innovation, you don't just want the good enough thing. You want

93
00:05:43,835 --> 00:05:47,509
the best thing. And it's in that area where when we really

94
00:05:47,509 --> 00:05:51,289
want the best solution, that our tendency to settle for good enough

95
00:05:51,669 --> 00:05:55,325
really hurts us. Because what teams do is they put

96
00:05:55,325 --> 00:05:59,165
in a prompt and they get a pretty good answer and they go, woah,

97
00:05:59,165 --> 00:06:02,400
I I was prepared to take an hour working on this but we kind of

98
00:06:02,400 --> 00:06:06,180
got pretty good in 5 minutes. You guys want to go get coffee? And everybody

99
00:06:06,480 --> 00:06:09,860
just gives up because they got good enough. And so that's,

100
00:06:10,365 --> 00:06:13,645
you know, I I think it really is a mindset thing. Forget the word prompt

101
00:06:13,645 --> 00:06:17,165
engineering. It's all it's it's self engineering. It's human

102
00:06:17,165 --> 00:06:20,610
engineering. And one of the best things that a human being can

103
00:06:20,610 --> 00:06:24,310
do is say to the AI why they don't like

104
00:06:24,449 --> 00:06:28,285
the answer the AI gave. Right? So take your expertise.

105
00:06:28,285 --> 00:06:31,965
Here, this is something that anyone is listening can do right now. Take something

106
00:06:31,965 --> 00:06:35,729
that you know you're an expert on. So for example, I'm an expert

107
00:06:35,729 --> 00:06:39,090
on customer insights or or low resolution prototyping and

108
00:06:39,090 --> 00:06:42,905
experimentation. Right? So I might say to the AI, can you give me a

109
00:06:42,905 --> 00:06:46,745
step by step guide for how to conduct an experiment? K. If I

110
00:06:46,745 --> 00:06:49,225
did that, we can do it right now live if we wanted to. But if

111
00:06:49,225 --> 00:06:52,520
I do that, it's gonna give me like, you know, the average of the Internet.

112
00:06:52,520 --> 00:06:55,320
Right? And it's gonna draw from a bunch of stuff that may be good, may

113
00:06:55,320 --> 00:06:58,920
be bad. By the way, I'm an expert, so the chances of my knowledge surpass

114
00:06:58,920 --> 00:07:02,625
it are reasonable, you know, at least. But because it's gonna give me

115
00:07:02,625 --> 00:07:05,425
the average of the Internet, it's probably gonna find some corners that I don't know

116
00:07:05,425 --> 00:07:09,260
about and it's probably gonna say some stuff that I vehemently disagree with. Well, where

117
00:07:09,260 --> 00:07:13,020
most people give up and I think because they want AI to not

118
00:07:13,020 --> 00:07:16,639
be that good is they look at the response and they go, see.

119
00:07:17,254 --> 00:07:20,694
It didn't even know that you're supposed to test for desirability and not

120
00:07:20,694 --> 00:07:24,294
feasibility. Right? Whatever. And then they say, that's why AI is no

121
00:07:24,294 --> 00:07:27,790
good. Well, Human Engineering, not Prompt Engineering.

122
00:07:27,930 --> 00:07:31,550
Human Engineering is to say, okay, human, tell the AI

123
00:07:32,330 --> 00:07:35,925
what you disagree with and why you disagree with it,

124
00:07:36,065 --> 00:07:39,125
and ask the AI to regenerate an answer

125
00:07:39,505 --> 00:07:42,965
given the following considerations and put in your critique.

126
00:07:44,180 --> 00:07:47,320
Most people if they do that, just even that one thought exercise,

127
00:07:48,020 --> 00:07:51,720
will be blown away. Yeah. You get an order of magnitude better response.

128
00:07:52,405 --> 00:07:56,004
Absolutely. Right? Absolutely. Because you're kind of focusing the cone

129
00:07:56,004 --> 00:07:59,384
of, you know, inquiry with your own

130
00:07:59,444 --> 00:08:03,180
expertise. And what people want is they I mean, no one would

131
00:08:03,180 --> 00:08:06,860
ever if you think about, like, AI like an MBA intern. Right? No

132
00:08:06,860 --> 00:08:10,655
one gets an intern from Harvard Business School, gives them 2

133
00:08:10,655 --> 00:08:14,414
sentences of instruction and then at the end of the summer says, man, their workout

134
00:08:14,414 --> 00:08:18,095
was no good. I didn't interact with them at all. I didn't give them any

135
00:08:18,095 --> 00:08:21,580
guidance. But for crying out loud, what's

136
00:08:21,580 --> 00:08:25,180
Harvard doing these days? Right? No. Nobody gives a

137
00:08:25,180 --> 00:08:29,020
human being 2 sentences of input and then critiques how bad of a job they

138
00:08:29,020 --> 00:08:32,684
did, right? And yet we open chat gpt, we give

139
00:08:32,684 --> 00:08:36,365
2 sentences of input, if that by the way, and then we go, See, it's

140
00:08:36,365 --> 00:08:40,059
not very good, well, work with it. Garbage in garbage

141
00:08:40,059 --> 00:08:42,860
out. It's as old as possible. And the reason most people don't I think most

142
00:08:42,860 --> 00:08:45,100
people don't wanna work with it is because they don't want it to be any

143
00:08:45,100 --> 00:08:48,755
good. Yeah. I could see that. Totally.

144
00:08:48,755 --> 00:08:52,215
And the people who do want it to be good will be unlocked and

145
00:08:52,275 --> 00:08:55,670
unleashed. But it requires not prompt engineering, but

146
00:08:55,670 --> 00:08:59,510
human copilot engineering. I do like the

147
00:08:59,510 --> 00:09:02,890
fact that a lot of these tools that are coming out are being called copilots,

148
00:09:03,110 --> 00:09:06,605
right? Because I think it shifts the focus away from

149
00:09:06,985 --> 00:09:10,445
AI isn't going to do it all. AI is not

150
00:09:10,585 --> 00:09:14,270
probably gonna take your job, right? But it's just an

151
00:09:14,270 --> 00:09:17,730
assistant. Right? It's it's to help you out where you may

152
00:09:18,190 --> 00:09:21,855
want a little bit of boost. I also think that I think what you you

153
00:09:21,855 --> 00:09:25,555
described is good enough factor is I think people see

154
00:09:26,015 --> 00:09:29,800
large language models and they they assume it's a search box only

155
00:09:29,800 --> 00:09:33,320
better. Yes. Well, and part of our

156
00:09:33,320 --> 00:09:36,805
challenge, you know, I was talking with a with a psychologist, David

157
00:09:36,805 --> 00:09:40,645
McCraney, who wrote How Minds Change. He's a he's a journalist, an author,

158
00:09:40,645 --> 00:09:44,200
a podcaster. He's a he's the host of You Are Not So Smart, which is

159
00:09:44,200 --> 00:09:47,960
all, you know, obsessed with cognitive bias, which I love. Mhmm. I love

160
00:09:47,960 --> 00:09:50,760
that podcast. And one of the things that David and I were talking about yeah.

161
00:09:50,760 --> 00:09:54,425
I can't remember the the name for the cognitive bias, but when we see

162
00:09:54,425 --> 00:09:58,205
something that we think we understand, we just track into

163
00:09:58,265 --> 00:10:01,980
our kind of typical neuro pathways. Right? So we

164
00:10:01,980 --> 00:10:05,500
see a text box and we go, oh, I've seen one of these

165
00:10:05,500 --> 00:10:09,185
before. This is like that. And we so and this

166
00:10:09,505 --> 00:10:13,185
this being generative AI is not like that. That being

167
00:10:13,185 --> 00:10:16,945
search. Generative AI is not search. But because of the

168
00:10:16,945 --> 00:10:20,529
kind of the the UI, we

169
00:10:20,529 --> 00:10:24,050
approach it like search. We go, okay, I want the answer. Just give me a

170
00:10:24,050 --> 00:10:27,795
list of links that I'm gonna click through and decide on. And we don't

171
00:10:27,795 --> 00:10:31,555
interrogate Google, we don't critique Google or any search

172
00:10:31,555 --> 00:10:34,855
engine, right? We don't say why we want it

173
00:10:35,155 --> 00:10:38,850
or state our intention, right? But if you start to do

174
00:10:38,850 --> 00:10:41,990
some of these fundamental kind of human

175
00:10:42,210 --> 00:10:45,795
conversational tactics, if you start treating it more like a

176
00:10:45,875 --> 00:10:49,495
person than like a search box, you get

177
00:10:49,875 --> 00:10:53,714
exponentially better results. But you're right, even

178
00:10:53,714 --> 00:10:57,120
the UI itself predisposes us to treat the

179
00:10:57,120 --> 00:11:00,880
technology and to think about the technology in a particular way and

180
00:11:00,880 --> 00:11:04,514
that is actually holding us back. Interesting. I noticed

181
00:11:04,514 --> 00:11:08,355
this in a completely random thing. I was getting,

182
00:11:09,394 --> 00:11:13,180
I was using DALL E to generate images, This is before Chat GPG had it

183
00:11:13,180 --> 00:11:16,540
in there. And I wanted to make a painting that looked like a Rembrandt painted

184
00:11:16,540 --> 00:11:19,600
a portrait of a dachshund. I know this is the most ridiculous thing.

185
00:11:20,345 --> 00:11:23,385
Right? So I wrote the prompt, I said, you know, painting of a dachshund in

186
00:11:23,385 --> 00:11:26,825
style of Rembrandt, and it produced something. It was okay. Right? It was

187
00:11:26,825 --> 00:11:30,210
good. But I was like, I wonder what if I asked

188
00:11:30,210 --> 00:11:34,050
ChatCpt to help me with this prompt? So I went over. Now

189
00:11:34,050 --> 00:11:36,834
I could do it all in 1 window. But I said, like, hey. What would

190
00:11:36,834 --> 00:11:39,714
what would you write for a prompt? Like, what would do that? And it came

191
00:11:39,714 --> 00:11:43,449
back with, I mean, a paragraph to what you said, 2 sentences. This thing came

192
00:11:43,449 --> 00:11:47,149
back with a paragraph. I mean, stuff that only art historians and art,

193
00:11:47,930 --> 00:11:51,689
students would really appreciate. You know, this type of paint, this style of

194
00:11:51,689 --> 00:11:55,525
brush, like, just stuff that I remember from art history class, but, like, you

195
00:11:55,525 --> 00:11:59,145
know, I only took that class because I had to type thing, you know?

196
00:11:59,765 --> 00:12:03,410
But but then I I pasted that prompt in there, and, oh,

197
00:12:03,410 --> 00:12:07,170
okay. It's it's it's an image. It's art. It it it's somewhat subjective, but

198
00:12:07,170 --> 00:12:11,005
the the result was so much better. Like, it was just day

199
00:12:11,005 --> 00:12:14,805
and night, and That's true. That has changed the

200
00:12:14,805 --> 00:12:18,400
way I think about, dare I say, prompt engineering. Right? Like, because you can

201
00:12:18,480 --> 00:12:21,440
because I gave a talk on prompt engineering and, like, you know, the magic of

202
00:12:21,440 --> 00:12:25,120
it, and I was like, you can actually have the models help you build out

203
00:12:25,120 --> 00:12:28,504
prompts. Yes. Well, that's that's the thing that people don't understand

204
00:12:28,504 --> 00:12:32,105
is, you know, I mean, I I interviewed the other

205
00:12:32,105 --> 00:12:35,620
day on a I've got a podcast called Beyond the Prompt, which is all about

206
00:12:36,100 --> 00:12:39,779
AI in organizations. And we've interviewed a bunch of amazing people.

207
00:12:39,779 --> 00:12:42,760
You have co founder of Typeform, CEO of Section,

208
00:12:43,345 --> 00:12:47,105
CEO of Every, the head architect at Instacart, a bunch of

209
00:12:47,105 --> 00:12:50,465
interesting people. And one of the folks we interviewed last week is a

210
00:12:50,465 --> 00:12:53,990
documentary filmmaker named Juan Carlos. And Juan Carlos has made some

211
00:12:53,990 --> 00:12:57,750
amazing documentaries. And he said he's always wanted to build

212
00:12:57,750 --> 00:13:01,430
an Ios application, but he's never had a developer and he's always seen that as

213
00:13:01,430 --> 00:13:05,205
kind of prohibited. He can't do it. And then he said when ChatGPT came

214
00:13:05,205 --> 00:13:08,565
out, he had the thought, could ChatGPT teach me how to

215
00:13:08,565 --> 00:13:12,370
code? And he built an

216
00:13:12,370 --> 00:13:16,050
Ios app by treating ChattGPT like his computer

217
00:13:16,050 --> 00:13:19,894
science TA. And he would go to the TA and ask for

218
00:13:19,894 --> 00:13:23,654
instructions. He got Chad GPT to teach him

219
00:13:23,654 --> 00:13:27,170
how to build an Ios app. Nice. You would

220
00:13:27,170 --> 00:13:30,472
never imagine doing that with a search engine, right? No. You would find it on.

221
00:13:30,472 --> 00:13:33,935
You would find it on. But you would just, and anytime he got stuck, You'd

222
00:13:33,935 --> 00:13:37,615
come back to the TA. Right? And you get more. But your

223
00:13:37,615 --> 00:13:41,135
point about people's minds being open, I think they have to be

224
00:13:41,135 --> 00:13:44,640
hearing examples like this. He literally went to

225
00:13:44,640 --> 00:13:48,480
JIGBT and said I would love to build an Ios app but I've never

226
00:13:48,480 --> 00:13:52,325
built anything. I don't have the first you know, sentence

227
00:13:52,704 --> 00:13:56,144
of ways to even describe it. If you were gonna ask a developer or if

228
00:13:56,144 --> 00:13:59,660
I wanted to ask a developer to do this, how would I even ask them?

229
00:13:59,660 --> 00:14:03,260
What do I need to describe? Tell me everything you need from me in

230
00:14:03,260 --> 00:14:06,585
order to tell me how to proceed. And he basically worked

231
00:14:07,145 --> 00:14:10,685
with it's almost reversed. We're used to being in the driver's seat.

232
00:14:10,905 --> 00:14:14,425
He basically told Chad GVT, you're in the driver's seat, please tell me what to

233
00:14:14,425 --> 00:14:18,150
do. I'll be your hands, you tell me what I need to do.

234
00:14:18,370 --> 00:14:22,210
And to me, that's just we have to start shifting paradigms. I'll

235
00:14:22,210 --> 00:14:25,905
give you another example. I've got a good friend who

236
00:14:25,905 --> 00:14:29,585
is considering a job transition. He lives on the East Coast,

237
00:14:29,585 --> 00:14:33,130
wants to move back to where his family is, and he got a

238
00:14:33,130 --> 00:14:36,730
job offer at a new firm. And he felt the job

239
00:14:36,730 --> 00:14:40,555
offer wasn't a great offer. His wife felt, we don't wanna screw

240
00:14:40,555 --> 00:14:43,755
this up. We wanna give back to family and we got a job. Just take

241
00:14:43,755 --> 00:14:47,490
the offer. And he he kind of confided in me, I

242
00:14:47,490 --> 00:14:50,530
feel like I could negotiate, but I don't want to mess things up. And I

243
00:14:50,530 --> 00:14:54,310
said, well, have you role played it with Chad GPT? And he said,

244
00:14:54,530 --> 00:14:58,255
what do you mean? I said, well, you can role play the conversation just

245
00:14:58,255 --> 00:15:01,215
to see how it would go. He said, but they don't know anything about the

246
00:15:01,215 --> 00:15:05,019
firm. I said, well, you can tell them. Ask ChadGpt, what do you need to

247
00:15:05,019 --> 00:15:07,839
know about the firm and what do you need to know about the hiring manager

248
00:15:08,300 --> 00:15:12,139
in order to believably play their role in a back

249
00:15:12,139 --> 00:15:15,885
and forth role play with with me. Interview me about the company and interview me

250
00:15:15,885 --> 00:15:19,725
about the person until you know enough to believably play

251
00:15:19,725 --> 00:15:22,980
their role and then do a 1 on 1 negotiation with me.

252
00:15:24,000 --> 00:15:27,680
Be observing the negotiation the whole time, and give me feedback not

253
00:15:27,680 --> 00:15:30,820
only as my counterparty, but also as a negotiation

254
00:15:31,120 --> 00:15:34,785
coach. That's brilliant. That's some sci fi

255
00:15:34,785 --> 00:15:38,545
stuff right there. Dude, he came back and he was like, what do I

256
00:15:38,545 --> 00:15:42,330
do now? That was mind blowing. I said, now, ask him

257
00:15:42,330 --> 00:15:45,530
to play your counterparty, but be a little bit more aggressive as the counterparty, a

258
00:15:45,530 --> 00:15:49,210
little bit less friendly. So he did that and he said, Jeremy, 2

259
00:15:49,210 --> 00:15:52,965
things I learned. 1, or actually 3 things. 1, I was

260
00:15:52,965 --> 00:15:56,665
missing my key point of leverage and ChatGPT helped me see it.

261
00:15:56,805 --> 00:16:00,339
2, I forgot my negotiating strategy in the

262
00:16:00,339 --> 00:16:03,380
in the heat of the moment, and chat g p t alerted me to that.

263
00:16:03,380 --> 00:16:07,220
Now I'm prepared. 3, I'm no longer dreading

264
00:16:07,220 --> 00:16:10,545
this negotiation. I know I can do it.

265
00:16:11,085 --> 00:16:14,685
Wow. And to me, it's like that's it's it's so different than

266
00:16:14,685 --> 00:16:18,339
saying, you know, portrait of a dash hound and, in remember

267
00:16:18,339 --> 00:16:21,500
it's like people are doing that and going, that's all I can do is like,

268
00:16:21,500 --> 00:16:24,714
you know, it can teach you how to build an Ios app. It taught me,

269
00:16:24,714 --> 00:16:27,435
I got you at GBT to teach me how to code Python so I could

270
00:16:27,435 --> 00:16:31,274
build my own chatbot using Python. I've literally never written a line of code

271
00:16:31,274 --> 00:16:35,050
in my entire life, right? It's our imaginations are the

272
00:16:35,050 --> 00:16:38,650
primary bottleneck here. And and part of the reason that our

273
00:16:38,650 --> 00:16:41,550
imagination is constrained is because we've been

274
00:16:42,065 --> 00:16:45,745
trained by search to interact

275
00:16:45,745 --> 00:16:49,505
with technology in a particular way. And what I think most people need is they

276
00:16:49,505 --> 00:16:53,060
need to hearing examples like this and they need to be getting in conversations with

277
00:16:53,060 --> 00:16:56,820
other people who are trying stuff and going, I can do that. Yeah,

278
00:16:56,820 --> 00:17:00,584
you could. I could do that. Yeah, you could. And you need to be having

279
00:17:00,584 --> 00:17:04,204
these kinds of conversations to stimulate your own thinking to then discover

280
00:17:04,825 --> 00:17:08,630
your own novel applications. No. That's brilliant. I

281
00:17:08,630 --> 00:17:12,150
mean, the whole negotiation thing is amazing. I've seen a lot of

282
00:17:12,150 --> 00:17:15,910
chatter online about people using it to, you

283
00:17:15,910 --> 00:17:19,734
know, in the job search aspect of it.

284
00:17:19,734 --> 00:17:22,395
Right? Like, here's the job description. Here's my current resume.

285
00:17:23,575 --> 00:17:27,380
Have at it, you know? Write Reno, write a cover letter that is

286
00:17:27,380 --> 00:17:31,220
gonna hit all these points and it'll do it. And, you know, but

287
00:17:31,220 --> 00:17:34,179
I mean, the whole idea of role playing. I mean, that's just brilliant. Like, I

288
00:17:34,179 --> 00:17:37,804
think I think the the the the the the $1,000,000

289
00:17:37,804 --> 00:17:41,345
statement there is our imagination

290
00:17:41,725 --> 00:17:45,550
is a limit, which is something that historically, when it comes to computers, I

291
00:17:45,550 --> 00:17:48,910
would say beyond the the the the the search

292
00:17:48,910 --> 00:17:52,130
interaction experience, we're not used to computers outthinking

293
00:17:52,350 --> 00:17:56,115
us. Yeah. Yeah. And I think that that that's gonna have

294
00:17:56,115 --> 00:17:59,654
some interesting, societal

295
00:17:59,875 --> 00:18:03,450
consequences. Right? Because I mean, I think what what freaked people out about Chat

296
00:18:03,450 --> 00:18:07,050
GPT was, you know, it looks like it's doing something

297
00:18:07,050 --> 00:18:10,825
creative, which is something that we had naively assumed, was

298
00:18:10,825 --> 00:18:14,585
something only humans can do. Mhmm. And I I I think you're right. I

299
00:18:14,585 --> 00:18:18,424
mean, I think this is not just a chat search only better, but this is

300
00:18:18,424 --> 00:18:22,200
definitely like a whole new type of computing. Yeah. I

301
00:18:22,200 --> 00:18:25,900
think it really does require a behavior modification. And

302
00:18:25,960 --> 00:18:29,625
what I I there there are kind of 2 big questions in my mind

303
00:18:29,625 --> 00:18:33,325
for organizations or for leaders who are thinking about deploying these technologies.

304
00:18:33,865 --> 00:18:37,625
1 is, what percentage of my workforce is comfortable

305
00:18:37,625 --> 00:18:41,370
with these tools? And by the way, right now, I mean, sentiment

306
00:18:41,450 --> 00:18:45,130
I read an Ernst and Young report that says 70% of people are afraid

307
00:18:45,130 --> 00:18:48,615
of AI. You know, it's like, when the when the predominant

308
00:18:48,755 --> 00:18:52,455
sentiment is fear, you're not in a position of kind of maximizing

309
00:18:52,595 --> 00:18:56,390
opportunity. Right? So Right. You so fear is gonna hold you back from

310
00:18:56,390 --> 00:19:00,230
that sense of comfort, confidence, etcetera. But then 2, so if you say so 1

311
00:19:00,230 --> 00:19:04,015
question is, what percent of our workforce is comfortable? And then

312
00:19:04,015 --> 00:19:06,915
2, how do I grow my conversation abilities?

313
00:19:08,015 --> 00:19:11,775
Nobody knows how to have a conversation right now with with HHPT or with

314
00:19:11,775 --> 00:19:15,600
any LLM. Many people have lost the art of having conversations with

315
00:19:15,600 --> 00:19:19,200
human beings, right? So, but you you really have

316
00:19:19,200 --> 00:19:22,985
to almost it's like becoming literate in a new language.

317
00:19:22,985 --> 00:19:26,505
We need AI literacy courses. We've actually developed, my

318
00:19:26,505 --> 00:19:30,309
partners and I, developed a conversational coach who gives

319
00:19:30,309 --> 00:19:34,150
daily drills that send you into ChatTPT with kind of a

320
00:19:34,150 --> 00:19:37,675
drill to build your conversational fluency. Because what we're finding

321
00:19:37,675 --> 00:19:41,515
is, folks just they don't have any imagination. Do you know that you could

322
00:19:41,515 --> 00:19:45,360
take ChatGPT, for example, and tell her what are your 5 favorite books

323
00:19:45,600 --> 00:19:48,500
and why they're your favorite books and ask for recommendations.

324
00:19:49,200 --> 00:19:53,040
It'll blow your mind. It'll give you recommendations that no human being's ever given

325
00:19:53,040 --> 00:19:56,654
you. Interesting. You could tell it you could tell it your, you

326
00:19:56,654 --> 00:20:00,174
know, 5 favorite quotes and ask for what

327
00:20:00,174 --> 00:20:03,770
are what are patterns here and what does it tell me about myself and my

328
00:20:03,770 --> 00:20:07,530
world view and what are my blind spots given these things that I'm drawn

329
00:20:07,530 --> 00:20:11,165
to. Right? You can take your journal entries and, you

330
00:20:11,165 --> 00:20:14,785
know, a particular difficult day that you've had recently.

331
00:20:15,005 --> 00:20:18,789
And then you can you can ask ChargeG PTE, can you tell me

332
00:20:18,789 --> 00:20:22,470
what are the mental models that are inhibiting my

333
00:20:22,470 --> 00:20:26,245
ability from seeing this situation clearly? And it will tell

334
00:20:26,245 --> 00:20:30,005
you, right? If it's That's wild. I'm just drawing on I

335
00:20:30,005 --> 00:20:33,044
mean, by the way, I'm just kind of a purveyor of these examples. They're all

336
00:20:33,044 --> 00:20:36,669
examples I've been hearing from people. But the point is, you can do so

337
00:20:36,669 --> 00:20:40,510
much more than you imagine. And right now, nobody's putting themselves

338
00:20:40,510 --> 00:20:44,325
or very few people even have kind of the the the wherewithal

339
00:20:44,385 --> 00:20:48,105
to say, I've gotta be hearing more of these examples. I wanna know my

340
00:20:48,225 --> 00:20:52,065
what my cognitive biases are. I wanna learn that new tool. I wanna try that

341
00:20:52,065 --> 00:20:55,780
thing. And the more examples you hear,

342
00:20:55,920 --> 00:20:59,680
the more your own imagination will be stimulated. Right? I mean, going back to idea

343
00:20:59,680 --> 00:21:03,434
flow or kind of my area of expertise which is innovation, creativity,

344
00:21:03,434 --> 00:21:06,795
etcetera. What we know cognitively is that the imagination is

345
00:21:06,795 --> 00:21:10,490
stimulated by unexpected inputs. So, you

346
00:21:10,490 --> 00:21:14,250
know, think back to Johannes Kepler gazing up in the night sky. Right? At

347
00:21:14,250 --> 00:21:18,090
that time, the predominant paradigm was, it's the firmament, meaning it is

348
00:21:18,090 --> 00:21:21,645
a fixed substance. Right? And Kepler sees a

349
00:21:21,645 --> 00:21:24,865
shooting star, and his first thought is,

350
00:21:25,325 --> 00:21:27,585
why isn't the firmament cracking?

351
00:21:28,940 --> 00:21:32,640
Right. And that is what led to heliocentricity.

352
00:21:33,740 --> 00:21:37,585
And, you know, the the total paradigm shift in the in

353
00:21:37,585 --> 00:21:41,025
the understanding of our place in the universe starts

354
00:21:41,025 --> 00:21:44,865
with a shooting star. Right? Unexpected inputs, sparks

355
00:21:44,865 --> 00:21:48,470
the imagination. And so that's that's that's a tactic

356
00:21:48,470 --> 00:21:52,309
that whether it's AI or anything else, putting yourself

357
00:21:52,309 --> 00:21:56,095
in the mindset of I need to be seeking unexpected input.

358
00:21:56,095 --> 00:21:59,794
Most people's lives are ordered to insulate

359
00:21:59,855 --> 00:22:03,680
and protect themselves from anything unexpected. And yet it's the

360
00:22:03,680 --> 00:22:07,440
unexpected which actually stimulates our imaginations and creates possibilities and

361
00:22:07,440 --> 00:22:11,265
opportunities for us and ideas. This is wild. I

362
00:22:11,265 --> 00:22:14,145
mean, like, I mean, one of the things that blew my mind was when they

363
00:22:14,145 --> 00:22:17,825
added the ability to create custom GPTs. Right? So I started

364
00:22:17,825 --> 00:22:21,620
tinkering with it, like, you know, if you listen to the show,

365
00:22:21,620 --> 00:22:25,460
we have a character named Bailey. So I kind of taught it, like, what would

366
00:22:25,460 --> 00:22:29,059
Bailey say? You know, this is the the idea for the character. This is kind

367
00:22:29,059 --> 00:22:32,535
of the her tone, and this is her personality that

368
00:22:32,535 --> 00:22:36,375
we've kind of defined. And for the last, I would say,

369
00:22:36,375 --> 00:22:40,060
15 episodes, that's actually what generates most of or all of the

370
00:22:40,060 --> 00:22:43,820
text that she says. Right? So it's kind of like I have my own

371
00:22:43,820 --> 00:22:47,660
private it's not Jarvis by any stretch of the imagination, like, you

372
00:22:47,660 --> 00:22:50,934
know, Iron Man, But I mean, it's kind of like, it's kind of like the,

373
00:22:50,934 --> 00:22:54,295
the, the, I have enough raw material there. I can

374
00:22:54,295 --> 00:22:57,835
pretend. Right? Cause the, the AI will say things like

375
00:22:58,230 --> 00:23:01,350
the, like, oh, yeah, that works. I like the way, I like the way she

376
00:23:01,350 --> 00:23:05,110
phrased that. And then I say she, because I mean, it's just funny. Like

377
00:23:05,110 --> 00:23:08,575
it's just, and and you know, there's ones where,

378
00:23:08,815 --> 00:23:12,575
there was a GPT I made where, you know, to help with motivation. It's like,

379
00:23:12,575 --> 00:23:16,230
you know, pretend you're Tony Robbins and you're trying to, like, motivate

380
00:23:16,230 --> 00:23:19,830
somebody to to do the best they can do. And, yeah, I've interacted with that.

381
00:23:20,150 --> 00:23:23,934
I'm impressed. I mean, it's just mind boggling what,

382
00:23:24,955 --> 00:23:28,635
it's mind boggling what these what this thing can do. And when as

383
00:23:28,635 --> 00:23:32,279
the engineer in me, I know this is just some kind of vector representation

384
00:23:32,419 --> 00:23:36,260
of language. It's a predictive model. Yeah. It's it's statistics. I

385
00:23:36,260 --> 00:23:39,539
think, you know, here's one thing I would say to listeners who may be dabbling,

386
00:23:39,539 --> 00:23:42,664
may be curious, whatever. If you get

387
00:23:44,005 --> 00:23:47,625
a bad output from a large language model,

388
00:23:48,404 --> 00:23:52,110
you need to start with the assumption it's because it was a

389
00:23:52,190 --> 00:23:55,870
you gave it a bad input. Right. And that's a really

390
00:23:55,870 --> 00:23:59,549
hard thing because we're we're used to saying if I get a bad output, it's

391
00:23:59,549 --> 00:24:03,295
because it's the model's no good. And where I really

392
00:24:03,295 --> 00:24:06,435
think we have to change some of our fundamental assumptions

393
00:24:06,735 --> 00:24:10,150
is the following: The problem

394
00:24:10,150 --> 00:24:13,530
isn't the technology, the problem is the user.

395
00:24:13,830 --> 00:24:17,315
And if we will take the burden of providing

396
00:24:17,375 --> 00:24:21,135
better input to the model, what we find is our mind starts

397
00:24:21,295 --> 00:24:24,515
I mean, I talked to someone the other day who said, almost daily,

398
00:24:25,269 --> 00:24:27,769
the AI does something that makes me giggle.

399
00:24:29,269 --> 00:24:33,029
And I think that that should be a goal. Like like, it's it's possible. I've

400
00:24:33,029 --> 00:24:36,745
had that experience. I mean, I'll I'll give you one example, Frank. We've built this

401
00:24:36,745 --> 00:24:40,585
series of drills as I mentioned to you. Right? That folks can connect their

402
00:24:40,585 --> 00:24:44,260
Slack or their Microsoft Teams to. And for an enterprise, they can get access for

403
00:24:44,260 --> 00:24:48,020
their employees where every individual employee gets drills on

404
00:24:48,020 --> 00:24:51,515
how to use generative AI better. Right? Well,

405
00:24:51,515 --> 00:24:55,355
we've only got a certain library of drills. Right? You know, and

406
00:24:55,355 --> 00:24:58,395
we're we're growing that. And every time we do a podcast, we learn something, we

407
00:24:58,475 --> 00:25:01,980
then we create a new drill. Right? We build that into all of the

408
00:25:01,980 --> 00:25:05,500
training information. Well, but there's still kind of you can still get to the end

409
00:25:05,500 --> 00:25:08,915
of the road. And I had this experience, and I just kind of pushed the

410
00:25:08,915 --> 00:25:12,434
coach to to just rapidly go through all the drills because I kind of wanted

411
00:25:12,434 --> 00:25:16,070
to see what happens when the sidewalk ends, like the old Shel Silverstein. Right?

412
00:25:16,149 --> 00:25:19,850
Right. What do we do whenever there's no more drills? And

413
00:25:20,630 --> 00:25:24,309
lo and behold, it suggested a drill that I had

414
00:25:24,309 --> 00:25:28,055
never thought of that was actually amazing. And I

415
00:25:28,055 --> 00:25:31,175
know I was and and I had that moment. Point being, I had the moment

416
00:25:31,175 --> 00:25:34,850
where I was giggling. Right? I think every single

417
00:25:34,909 --> 00:25:38,269
human being should seek for a moment

418
00:25:38,269 --> 00:25:41,409
where generative AI makes you giggle with delight.

419
00:25:41,789 --> 00:25:45,555
Right. Or makes you sit down in your chair and smack your head far

420
00:25:45,555 --> 00:25:49,235
ahead and go, wow. Yes. You know, I always think of Keanu

421
00:25:49,235 --> 00:25:53,040
Reeves in in the the first major movie. Woah. Like, I have a

422
00:25:53,040 --> 00:25:56,800
lot of those moments where I'm, like, wait, what? You

423
00:25:56,800 --> 00:26:00,020
know, like, wow. It's it's it's an impressive,

424
00:26:00,545 --> 00:26:04,225
and and and again, like, I think maybe being an engineer, where I

425
00:26:04,225 --> 00:26:08,065
see it is a cognitive bias in itself, right? I see it as

426
00:26:08,065 --> 00:26:11,730
some kind of vector representation of language, as being run over by some

427
00:26:11,730 --> 00:26:15,110
kind of statistical processing. But clearly,

428
00:26:15,410 --> 00:26:18,945
the sum of the parts is is more

429
00:26:19,405 --> 00:26:22,845
than the whole is more than the sum. I don't know, like, it's just one

430
00:26:22,845 --> 00:26:26,125
of those things where it makes me stop and ponder, like, what what have we

431
00:26:26,125 --> 00:26:29,830
built here? Like, what

432
00:26:30,630 --> 00:26:34,230
and what what's it doing that we can't see? Or what what what else is

433
00:26:34,230 --> 00:26:36,875
beyond there at all? It it opens up a sense, for lack of a term,

434
00:26:36,875 --> 00:26:40,635
like a sense of wonder. Like, you know, what else could I ask it? Right?

435
00:26:40,635 --> 00:26:44,315
Right. Right. And I think that that's everybody needs to get to

436
00:26:44,315 --> 00:26:47,860
that moment. And right now, too many people are sitting on the sidelines rather

437
00:26:47,860 --> 00:26:51,620
than you know, one one thing that everybody can do sorry. I wanna

438
00:26:51,620 --> 00:26:55,365
say 2 things. 1, to your point, I heard Sam Altman the other

439
00:26:55,365 --> 00:26:59,205
day. Someone asked, well, how's OpenAI gonna make money? And he said, well, we'll just

440
00:26:59,205 --> 00:27:02,620
ask the AI. We thought that was great.

441
00:27:03,580 --> 00:27:06,220
But the other thing I was gonna say is if folks are seeking kind of

442
00:27:06,220 --> 00:27:09,705
one of these personal epiphanies, here's the first. Well, the 1st drill

443
00:27:10,004 --> 00:27:13,684
in the kind of coach architecture is download Chi

444
00:27:13,684 --> 00:27:17,330
ChiPT's app and put it on your home screen. You're not going to use something

445
00:27:17,330 --> 00:27:20,890
that you don't see regularly. Right? So put on your home screen, that's kind of,

446
00:27:20,890 --> 00:27:24,575
you know, that's assignment number 1. And then assignment number 2 is

447
00:27:24,575 --> 00:27:28,015
think of an emotional decision you're trying to make right now in your

448
00:27:28,015 --> 00:27:31,615
life. Just personally, not related to work. I mean, it could be, I guess.

449
00:27:31,615 --> 00:27:35,110
But it has to be emotional. The kind of thing that you would ordinarily

450
00:27:35,649 --> 00:27:39,250
talk to a human being about. It can be

451
00:27:39,250 --> 00:27:42,605
anything. For me, like, I recently, I

452
00:27:42,605 --> 00:27:46,445
was wondering whether I should move my family. We had an opportunity to

453
00:27:46,445 --> 00:27:49,965
move. And I didn't really know how to think about it. What the what like,

454
00:27:49,965 --> 00:27:53,270
how to, to weigh the pros and cons. And so I actually reached out to

455
00:27:53,270 --> 00:27:57,110
a number of mentors and folks who I trust to talk about that. That

456
00:27:57,110 --> 00:28:00,395
kind of a topic. I I have a friend who told me he did this

457
00:28:00,395 --> 00:28:04,075
with his grandma and she asked the question, when is the time to

458
00:28:04,075 --> 00:28:07,859
move into assisted living? Right. That's a tough one. Right.

459
00:28:07,859 --> 00:28:11,580
So, yeah. So it's big questions like that. Right? Take a question like that

460
00:28:11,580 --> 00:28:15,395
that you'd ordinarily ask a trusted human being and go to

461
00:28:15,395 --> 00:28:18,995
Chattopty and say, I'd like to ask you about, for

462
00:28:18,995 --> 00:28:22,230
me, whether I should move my family to a new home.

463
00:28:22,790 --> 00:28:26,490
Before I do, would you ask me 4 or 5 questions

464
00:28:26,870 --> 00:28:30,534
so that you can better understand where I am in my

465
00:28:30,534 --> 00:28:33,595
life so that your advice can be tailored to my situation.

466
00:28:34,215 --> 00:28:37,255
And then oh, and do it 1 at a time because I'm a human and

467
00:28:37,255 --> 00:28:40,620
I can't handle more than 1 question at a time. Right? Well, then what

468
00:28:40,620 --> 00:28:44,300
ChatGPT does is it starts asking questions. Well, tell me about your current living situation.

469
00:28:44,300 --> 00:28:47,260
Well, tell me about this new place. Tell me you know, and it will ask

470
00:28:47,260 --> 00:28:51,095
3 or 4 questions and then it'll give

471
00:28:51,155 --> 00:28:54,915
amazing advice that you go, wow. That's I

472
00:28:54,915 --> 00:28:58,260
mean, you know, my friend who did this with his grandma said

473
00:28:59,600 --> 00:29:03,140
she told him this is genuinely new

474
00:29:03,280 --> 00:29:07,075
information and perspective that I hadn't considered. And all

475
00:29:07,075 --> 00:29:10,915
it took was me being 1, being willing to ask a vulnerable question, and

476
00:29:10,915 --> 00:29:14,675
2, being willing to answer a handful of questions that the AI asked

477
00:29:14,675 --> 00:29:18,440
me before I'm open to receiving input. Right?

478
00:29:18,980 --> 00:29:22,600
And it's it's it's really so taking a personal

479
00:29:23,715 --> 00:29:27,075
kind of, emotional decision to the AI is a really great

480
00:29:27,075 --> 00:29:30,835
way to stimulate one of these epiphanies. I feel like

481
00:29:30,835 --> 00:29:34,340
once you have one of these, kind of, personal epiphanies, you're off to the

482
00:29:34,340 --> 00:29:38,120
races. My friend told me his grandma's like all of a sudden going, you know,

483
00:29:38,900 --> 00:29:42,544
at the family holiday party, We're out of cream of mushroom

484
00:29:42,605 --> 00:29:46,225
soup, for the green bean casserole. Could Jaijibiti

485
00:29:46,445 --> 00:29:50,030
give me a replacement for cream of mushroom soup? Like, in what world

486
00:29:50,030 --> 00:29:53,870
does the 90 year old grandma ask that kind of question of Chad

487
00:29:53,870 --> 00:29:57,390
GPT? It's the world in which she had already talked

488
00:29:57,390 --> 00:30:01,164
about whether she should move into Assisted Living. Right? And she's had that

489
00:30:01,164 --> 00:30:05,005
personal epiphany. I feel like in a lot of companies, the company is asking

490
00:30:05,005 --> 00:30:08,650
employees, what can Generative AI do for our business?

491
00:30:09,350 --> 00:30:12,970
And most employees can't answer the question because they don't know what Generative

492
00:30:13,110 --> 00:30:16,645
AI can do. Right. So how can they know what it can do for the

493
00:30:16,645 --> 00:30:19,465
business? And so you've had some of these personal experiences.

494
00:30:20,085 --> 00:30:23,525
Don't don't be thinking about the business. Think about it it seems

495
00:30:23,525 --> 00:30:27,190
paradoxical, but I find that you have to explore the kind

496
00:30:27,190 --> 00:30:30,710
of possibility space individually, and then you start

497
00:30:30,710 --> 00:30:33,610
sparking just like grandma on the kid. Could it

498
00:30:34,315 --> 00:30:37,915
recommend a substitute for cream of mushroom soup? Well, yeah, it

499
00:30:37,915 --> 00:30:41,355
could. Could it but you have to have that personal epiphany

500
00:30:41,355 --> 00:30:44,700
first. Right. Because it's not something you would think about

501
00:30:45,000 --> 00:30:48,780
when you think about computers. Computers have historically been seen as very very

502
00:30:49,160 --> 00:30:52,925
logical, very emotional. Right? I was watching an old episode of,

503
00:30:53,245 --> 00:30:56,945
Star Trek The Next Generation, and there was 1 episode where,

504
00:30:58,240 --> 00:31:01,919
Data was asked to be and there was a line in

505
00:31:01,919 --> 00:31:05,760
there that kinda stuck me as funny because when I remember watching this when it

506
00:31:05,760 --> 00:31:09,225
originally aired, but I hear it now, it kinda makes me laugh, where he

507
00:31:09,225 --> 00:31:12,985
says, Data can be the judge of this because he's an artificial intelligence,

508
00:31:12,985 --> 00:31:16,365
and artificial intelligence have no biases, and will act unemotionally.

509
00:31:17,770 --> 00:31:20,429
And I'm kind of like, wow, that didn't age well.

510
00:31:21,530 --> 00:31:25,309
Yeah. Yeah. You know, it's right now, it's limited by our biases.

511
00:31:25,774 --> 00:31:28,975
Right. And that's the problem is we have a lot of to your point, we

512
00:31:28,975 --> 00:31:31,855
have a lot of bias. Even what you said, right, about being an engineer and

513
00:31:31,855 --> 00:31:35,290
thinking it's just a predictive model. Right. That bias limits your

514
00:31:35,290 --> 00:31:39,130
own you can't imagine what quote just a predictive model can

515
00:31:39,130 --> 00:31:42,845
actually do. Right? 100% as long as you think about it As

516
00:31:42,845 --> 00:31:45,525
as long as you think about it's just a predictive model or it's just an

517
00:31:45,525 --> 00:31:49,125
AI and so it doesn't have bias, what you fail to realize is the bias

518
00:31:49,125 --> 00:31:52,669
you bring as the co pilot shapes the entire

519
00:31:52,730 --> 00:31:56,270
trajectory of the thing. It's like a giant chameleon, isn't it?

520
00:31:56,490 --> 00:31:59,049
It is. Yeah. That's a good way to put it. That's a great way to

521
00:31:59,049 --> 00:32:02,505
put it. And the more into that end, or using that

522
00:32:02,505 --> 00:32:06,345
metaphor, the more environments that you place it in, the

523
00:32:06,345 --> 00:32:09,580
more you can appreciate its complexity and range,

524
00:32:09,880 --> 00:32:13,240
etcetera. Yeah. And this isn't we've we've used

525
00:32:13,240 --> 00:32:16,275
ChatGPT as an example, but like, so there was a,

526
00:32:17,155 --> 00:32:20,915
somebody at work had built a, basically completely open

527
00:32:20,915 --> 00:32:24,295
source language model based on documentation for a product.

528
00:32:25,080 --> 00:32:28,520
And I had meant to ask the chatbot, how do you

529
00:32:28,520 --> 00:32:31,720
connect it? How do you connect this cluster to a GPU? Or how do you

530
00:32:31,720 --> 00:32:34,904
add GPU as a resource? But what I I meant to say, how do you

531
00:32:34,904 --> 00:32:37,945
make a cluster with GPU? But I ended up typing, how do you make a

532
00:32:37,945 --> 00:32:41,705
GPU? Right? And what was what was

533
00:32:41,705 --> 00:32:44,680
interesting was I've written chatbots, you know, pre,

534
00:32:45,540 --> 00:32:49,140
transformer, and it would basically say, don't understand the question or you can't make a

535
00:32:49,140 --> 00:32:52,200
GPU or get confused. This basically gave me

536
00:32:53,115 --> 00:32:56,794
an entire 2 sentences of hey, very nicely, by the

537
00:32:56,794 --> 00:32:59,355
way, I might add, where it said, I'm sorry, but you feel like I can't

538
00:32:59,355 --> 00:33:03,130
really create a GPU for you. GPU's are hardware. And it went through and

539
00:33:03,130 --> 00:33:06,750
explained, like, the manufacturing process of a GPU.

540
00:33:07,370 --> 00:33:10,965
Wow. I I thought that was funny. And I screenshotted to the

541
00:33:10,965 --> 00:33:14,725
guy who made it because for for me, it was a typo.

542
00:33:14,725 --> 00:33:18,325
But from you know, I thought it just it was beautiful the way it

543
00:33:18,325 --> 00:33:22,040
answered it. Right? Yeah. That's great. That's Which was

544
00:33:22,440 --> 00:33:26,120
it it made me laugh. And, I don't think

545
00:33:26,120 --> 00:33:29,755
people realize that. Like, it it just because he didn't

546
00:33:29,755 --> 00:33:33,515
program it for that. He basically, you know, took a base

547
00:33:33,515 --> 00:33:37,260
model and and and, you know, sent it all our docs as kind of

548
00:33:37,260 --> 00:33:40,320
a it wasn't quite rag, but close enough.

549
00:33:41,660 --> 00:33:44,700
But it was just funny, like but it was nice about it too, which I

550
00:33:44,700 --> 00:33:48,285
thought was also interesting. But it

551
00:33:48,285 --> 00:33:51,885
was the kind of question you would get from, like, like, you know, someone who's

552
00:33:51,885 --> 00:33:55,659
not in technology. Can you make me a GPU? I don't know. I just

553
00:33:55,860 --> 00:33:59,620
I for me, that that every time I interact with this, it always moves the

554
00:33:59,620 --> 00:34:03,404
bar on, you know, where my bias was. Like, you know. Well, it's and that's

555
00:34:03,404 --> 00:34:07,164
a good that's a good thing to mention is it's it's a

556
00:34:07,164 --> 00:34:10,530
function of reps and exposure. And right now, if you

557
00:34:10,530 --> 00:34:14,290
find your imagination isn't sparked, put in a little bit more time. And

558
00:34:14,290 --> 00:34:17,875
this is where you kinda have to take on faith, but just give it a

559
00:34:17,875 --> 00:34:21,234
try. You know, to spend a few hours a week. You know, if you haven't

560
00:34:21,234 --> 00:34:24,690
had minimum of 10 hours in ChatGPT, you have

561
00:34:24,690 --> 00:34:28,070
no basis for dismissing the technology. None whatsoever.

562
00:34:28,369 --> 00:34:31,915
100%. You don't have, you know, 5 I'm looking at just at my

563
00:34:31,915 --> 00:34:35,675
Chrome browser right now. I have 5 windows ChatGPT windows open right

564
00:34:35,675 --> 00:34:39,469
now. If you don't have at least 5 windows open right now, you have it,

565
00:34:40,030 --> 00:34:43,090
that's a really kind of funny, somewhat

566
00:34:44,109 --> 00:34:47,949
binary question. How many tabs of Chat GPT do you

567
00:34:47,949 --> 00:34:51,304
have open? Usually, it's 0 or 15.

568
00:34:51,844 --> 00:34:55,605
That's right. That's right. And if you're in the zero camp, that's

569
00:34:55,605 --> 00:34:58,505
fine, but you have to go, why are really smart people

570
00:34:59,930 --> 00:35:03,710
running 15 tabs of this thing right now? Like, what am I missing?

571
00:35:03,850 --> 00:35:07,694
And how could I be this is an Ironman suit. Right? How

572
00:35:07,694 --> 00:35:11,375
could I be amplified? How am I not being amplified that I could be?

573
00:35:11,375 --> 00:35:14,895
Right? And taking that a little bit of the burden of proof and placing it

574
00:35:14,895 --> 00:35:18,710
on yourself, I think is, again, that's not something that

575
00:35:18,710 --> 00:35:22,470
we are that we are apt to do as human beings. And

576
00:35:22,470 --> 00:35:25,935
yet those who have done it have they're experiencing incredible

577
00:35:26,235 --> 00:35:29,995
benefits, incredible, delight, to your point. There's a

578
00:35:29,995 --> 00:35:32,955
lot of delight to be had, but you've got to kind of put yourself in

579
00:35:32,955 --> 00:35:36,540
that position. And I I've used it, I'll admit I've used it where I'll I'll

580
00:35:36,540 --> 00:35:40,060
write something in both my personal and professional life, and I'm like, well, can you

581
00:35:40,060 --> 00:35:43,535
make that nicer? Can you make it more persuasive? That's

582
00:35:43,535 --> 00:35:46,835
great. And it does an awesome job of that, you know?

583
00:35:49,050 --> 00:35:52,330
I'm just I'm continually amazed by it, you know. But

584
00:35:52,970 --> 00:35:56,615
and and I don't I keep it to a couple of tabs. If

585
00:35:56,615 --> 00:36:00,075
you're actively, like, having it generate text Mhmm.

586
00:36:00,215 --> 00:36:03,275
Doesn't it lock you out of the other ones too, or is that just

587
00:36:03,840 --> 00:36:07,120
you can if you had the real okay. Now now this is gonna be mind

588
00:36:07,120 --> 00:36:10,720
blowing. Cool. Yeah. No. You you know what I'll do too. I mean, and even

589
00:36:10,720 --> 00:36:14,305
for, like, demos with this Right. With this coach with this, you know, kind of

590
00:36:14,305 --> 00:36:17,984
drill coach, I'll I'll say, you know, I'll be

591
00:36:17,984 --> 00:36:21,505
in the tab on my Chrome and I'll be saying, you know, I'll be kinda

592
00:36:21,505 --> 00:36:25,069
giving instructions and I say I wanna go to voice mode now I'll pick up

593
00:36:25,069 --> 00:36:28,829
my device and I'll go into that chat so it's got all the

594
00:36:28,829 --> 00:36:32,245
context of that chat and then I'll turn it on to voice mode

595
00:36:32,705 --> 00:36:36,485
and then and and so now the user is kinda watching me with the camera.

596
00:36:36,705 --> 00:36:40,359
Well, then I wanna go back into the chat after the voicemail because I

597
00:36:40,359 --> 00:36:44,119
wanted to evaluate the conversation and I just reload the page and now

598
00:36:44,119 --> 00:36:47,319
all of a sudden everything I said that they just watched me say and everything

599
00:36:47,319 --> 00:36:51,085
that ChatGPT sent back to me is now on the screen. That's wild.

600
00:36:51,085 --> 00:36:54,725
I have to try the app in the voice mode. You have to. No. That's

601
00:36:54,925 --> 00:36:58,610
it's, you know, I mean, that's another activity. You know, again, if if folks

602
00:36:58,610 --> 00:37:01,650
wanna learn more about this research, because there's a lot of research behind this, you

603
00:37:01,650 --> 00:37:05,010
can go to how to fix it dot ai. That's a simple website that we

604
00:37:05,010 --> 00:37:08,685
set up. Because fix it is the model that we've put forth, f

605
00:37:08,685 --> 00:37:12,525
I x I t. But and we can talk to that if you

606
00:37:12,525 --> 00:37:15,990
want to. But If if if you if you go to how to fix it

607
00:37:15,990 --> 00:37:19,750
dot ai, you can download our research paper, all that stuff.

608
00:37:19,750 --> 00:37:23,270
It's all there. But one of the one of the drills that we offer in

609
00:37:23,270 --> 00:37:26,295
this drill coach is after a phone call,

610
00:37:28,035 --> 00:37:31,715
just do a verbal vomit into chat g p t. Open it up on your

611
00:37:31,715 --> 00:37:35,230
device, on your, you know, on your on your mobile device,

612
00:37:35,770 --> 00:37:38,490
put it in voice mode, and then, you know, you and I, Frank, we're talking

613
00:37:38,490 --> 00:37:41,455
right now. I might go in after and say, hey, I had a great, you

614
00:37:41,455 --> 00:37:43,855
know, lit literally. Okay. Here, I'll do it right now. Just so you can see

615
00:37:43,855 --> 00:37:47,295
how it would work. It's it's this simple. So I'm opening

616
00:37:47,295 --> 00:37:51,040
TagTpT up on my phone for people who, you know, can see. I don't

617
00:37:51,040 --> 00:37:53,520
know. And now I'm gonna go into the and I'm just gonna hit the whisper

618
00:37:53,520 --> 00:37:57,200
button, which kind of gives you voice mode. Not the headphones. I don't like back

619
00:37:57,200 --> 00:38:00,395
and forth. I mean, you can do that, but just in the text box, you

620
00:38:00,395 --> 00:38:04,234
hit that. I'll say, hey. So I'm talking with Frank right now on his

621
00:38:04,234 --> 00:38:07,674
podcast, and I wanna send a quick thank you note. Let him know how much

622
00:38:07,674 --> 00:38:11,450
I appreciate not only his humility, but also how he can share

623
00:38:11,450 --> 00:38:15,290
personal examples. It really felt like a back and forth and like a conversation. And

624
00:38:15,290 --> 00:38:18,765
so many podcasts feel like, you know, bland and,

625
00:38:19,465 --> 00:38:22,665
like they're just rote. They're asking the same old questions. I really felt like he

626
00:38:22,665 --> 00:38:26,460
was very engaged and thoughtful and it meant a lot to me. So would you

627
00:38:26,460 --> 00:38:30,059
mind to make a quick memo? Oh, yeah. And don't forget to send them a

628
00:38:30,059 --> 00:38:33,740
link to how to fix it dot ai. If you don't mind,

629
00:38:33,740 --> 00:38:37,485
just compose a quick memo that I could read, and copy paste this

630
00:38:37,485 --> 00:38:41,265
in there. So I just noticed how unstructured

631
00:38:41,325 --> 00:38:45,020
I was. Right? Oh, yeah. It was very chaotic. Very normal. Vomiting. Right? Well, I

632
00:38:45,020 --> 00:38:48,720
just did I just converted all that to text. I hit the up arrow

633
00:38:49,180 --> 00:38:52,975
and instantaneously, sure thing, here's a heartfelt memo for Frank. Hey, Frank.

634
00:38:52,975 --> 00:38:56,415
Just wanted to drop you a quick note to express my sincere gratitude for having

635
00:38:56,415 --> 00:39:00,050
me on your podcast. Your humility and willingness to share personal stories

636
00:39:00,050 --> 00:39:03,730
truly transformed our chat into a dynamic conversation, far from the

637
00:39:03,730 --> 00:39:07,570
conventional scripted exchanges that are all too common. It was refreshing to

638
00:39:07,570 --> 00:39:11,315
engage in a dialogue that felt both meaningful and genuine. Okay. I'm

639
00:39:11,315 --> 00:39:14,115
also including a link to how to fix it.ai as I believe it could offer

640
00:39:14,115 --> 00:39:17,640
additional value to our conversation and your audience. Thanks again for the opportunity.

641
00:39:17,640 --> 00:39:21,400
Right? But the point is now watch this. Now you go. It

642
00:39:21,400 --> 00:39:25,180
feels a little bit emotional. Can you make it a little bit more

643
00:39:25,815 --> 00:39:29,655
impersonal and also half the length? Because I'm afraid you may not be able

644
00:39:29,655 --> 00:39:33,470
to read it through all this. Right? And

645
00:39:33,470 --> 00:39:37,230
then again, it's, you know, it's absolutely, let's streamline it.

646
00:39:37,230 --> 00:39:40,795
Hey, Frank. Thanks for our engaging conversation on the podcast. Appreciate your thoughtful approach

647
00:39:40,954 --> 00:39:44,075
you shared insights, making it more than just the usual q and a. Here's a

648
00:39:44,075 --> 00:39:46,954
link to how to fix it dot ai that might interest you and your listeners.

649
00:39:46,954 --> 00:39:50,174
Cheers, Jeremy. That is just

650
00:39:50,680 --> 00:39:54,440
But these seem like it's easy. Like whereas whereas I might forget to

651
00:39:54,440 --> 00:39:58,280
do that, right? I might never send you it. I'll send you this just

652
00:39:58,280 --> 00:40:01,724
for your fun, right? But the point is there's so many things that just slipped

653
00:40:01,724 --> 00:40:05,345
through the cracks because like we're we're moving well. You know, right after this

654
00:40:05,885 --> 00:40:09,710
podcast I wanna go on a run. Typically, I'm stretching for the run. And

655
00:40:09,710 --> 00:40:13,410
now, Chad GPT has transformed my stretch time from kind of mindless

656
00:40:13,789 --> 00:40:17,425
to I can, you know, just unload. I mean, maybe

657
00:40:17,425 --> 00:40:20,065
sometimes I have like 3 or 4 sales calls in the morning or I've or

658
00:40:20,065 --> 00:40:23,810
I've got office hours, I've got meetings with students, whatever it might be. But

659
00:40:23,810 --> 00:40:26,790
I can just do like a verbal vomit literally

660
00:40:27,650 --> 00:40:31,410
and then ask JGPT to synthesize it for me. Send me a note to myself

661
00:40:31,410 --> 00:40:35,125
that I don't forget after I go on a run. Right? These 5 things

662
00:40:35,125 --> 00:40:38,805
I need to do. Right? And the point is, it's it's just

663
00:40:38,805 --> 00:40:42,550
about learning. I can do that? Yeah. You can do that.

664
00:40:42,550 --> 00:40:45,430
Right? And that's what we're trying to do with our drill coach is just give

665
00:40:45,430 --> 00:40:49,255
people a bunch of things that, yeah, you can do that. Not because that's

666
00:40:49,255 --> 00:40:52,555
the end point, but because it's a starting point for their own imagination.

667
00:40:53,335 --> 00:40:56,455
Yeah. I mean, that's imp I mean, that's mind boggling because, you know, there's a

668
00:40:56,455 --> 00:41:00,280
lot of, I guess, brain spillage you could capture with this and kind of, you

669
00:41:00,280 --> 00:41:04,040
know, move it forward because that happens to me all the time. I can't wait

670
00:41:04,040 --> 00:41:07,395
to see if this is gonna be integrated with Apple Auto or Android Android Auto

671
00:41:07,395 --> 00:41:11,154
or Apple Car because that would be epic. Because I get my best

672
00:41:11,154 --> 00:41:14,970
ideas when I'm driving. So so tell me about this

673
00:41:14,970 --> 00:41:18,170
FIXIT framework. Because whenever I hear FIXIT, I have a 1 year old and I

674
00:41:18,170 --> 00:41:21,994
think Bob the Builder. That's hysterical. That's hysterical.

675
00:41:21,994 --> 00:41:25,835
Well, FIXIT is just the acronym. Right? FIXIT. And it's

676
00:41:25,835 --> 00:41:29,675
basically it's we think the way we converse with AI is broken. So here's

677
00:41:29,675 --> 00:41:33,430
how to fix it. F is to have a focused question. So really be

678
00:41:33,510 --> 00:41:37,349
you know, it's not how do I create a Scratchy prototype. It's I'm trying to

679
00:41:37,349 --> 00:41:40,865
create a chatbot that teaches people how to have a conversation with

680
00:41:40,865 --> 00:41:44,625
AI. Right now, all of my users are doing this annoying thing and I

681
00:41:44,625 --> 00:41:47,990
don't know what's happening. I'm trying to, increase

682
00:41:48,609 --> 00:41:52,369
how often they return to finish a lesson rather than leaving

683
00:41:52,369 --> 00:41:56,125
and having me re engage. So F is a focused question.

684
00:41:56,125 --> 00:41:59,265
That's an example of a focused question. I is individually

685
00:41:59,724 --> 00:42:03,425
ideate. Before you brainstorm with IGBT or with a team

686
00:42:03,720 --> 00:42:07,320
think for yourself. What do I think about this? Too often people come with like

687
00:42:07,320 --> 00:42:10,460
a like, they're thoughtless. And the thing is thoughtlessness

688
00:42:10,840 --> 00:42:14,515
inhibits the context you can provide to GPT. That's what the X is

689
00:42:14,515 --> 00:42:17,734
for, FIX. X is give, provide context.

690
00:42:18,035 --> 00:42:21,255
Upload documents, here's transcripts from previous interactions.

691
00:42:22,140 --> 00:42:25,980
Here's our one pager for the Drill Coach and how we've been

692
00:42:25,980 --> 00:42:29,660
describing it. Here's a video of a user navigating for the

693
00:42:29,660 --> 00:42:33,275
1st time. Right? Whatever it is, give minimum 400 characters,

694
00:42:33,494 --> 00:42:37,255
provide sufficient context for the AI. Next I, so f

695
00:42:37,255 --> 00:42:40,395
I x I, this is interact iteratively.

696
00:42:41,110 --> 00:42:44,710
So you're having a back and forth whatever chat gpt gives you, ask it to

697
00:42:44,710 --> 00:42:48,470
regenerate. Critique the response. I don't get this. This doesn't make sense. I

698
00:42:48,470 --> 00:42:52,295
never would've thought that, right? Many times you're going to get junk

699
00:42:52,295 --> 00:42:55,915
output that's fine. Iterate, iteratively interact.

700
00:42:56,295 --> 00:42:59,789
And then T is team incubation. So once you get

701
00:42:59,789 --> 00:43:03,549
input from JGPT take it to the team and think about how do we

702
00:43:03,549 --> 00:43:07,105
commission a series of experiments to test which of these

703
00:43:07,105 --> 00:43:10,725
ideas actually solves the problem in the best way. Right? And so,

704
00:43:12,145 --> 00:43:15,799
I had a guest on my podcast describe generative AI as like an

705
00:43:15,799 --> 00:43:19,160
electric bike for the mind, which I love. Right? It's not an

706
00:43:19,160 --> 00:43:22,665
autonomous vehicle. It's not gonna do everything in parallel park. An

707
00:43:22,665 --> 00:43:26,185
electric bike, you can climb bigger cognitive hills, you can

708
00:43:26,185 --> 00:43:29,945
climb greater cognitive distances, you still have to steer the thing. You've

709
00:43:29,945 --> 00:43:33,069
got to be aware of traffic. You've got to be watching the lights. You've got

710
00:43:33,369 --> 00:43:36,809
to park the car. Yeah and walk through the threshold of your

711
00:43:36,809 --> 00:43:40,270
destination, right? And so bringing it back to the team

712
00:43:40,875 --> 00:43:44,575
and having a conversation with the team is an essential part of maximizing

713
00:43:44,875 --> 00:43:48,640
the output of AI. Right? So FIXIT, we've seen that

714
00:43:48,640 --> 00:43:52,340
folks who really provide a focus problem, individually ideate,

715
00:43:52,560 --> 00:43:56,160
provide sufficient context, interact iteratively with the

716
00:43:56,160 --> 00:44:00,005
language model, and then include their team in the incubation process,

717
00:44:00,465 --> 00:44:04,005
those folks dramatically outperform folks who just

718
00:44:04,145 --> 00:44:07,205
interact with the with the LLM like it's an oracle.

719
00:44:08,579 --> 00:44:11,700
I mean, that's very well said. I think that sums it all up, which I

720
00:44:11,700 --> 00:44:14,820
I like to fix it. I have the little Bob the Builder theme song in

721
00:44:14,820 --> 00:44:18,555
my head. I won't sing or it for multiple reasons, not the

722
00:44:18,555 --> 00:44:21,215
least of which is copyright. Come on. But,

723
00:44:22,395 --> 00:44:24,790
but what's, I mean, it's just interesting, though, like,

724
00:44:27,110 --> 00:44:29,990
it's so simple in a lot of ways. Like, this is this is but but,

725
00:44:29,990 --> 00:44:33,775
like, it it all makes sense. Right? You know? And and here's the

726
00:44:33,775 --> 00:44:37,615
thing, maybe this is the engineer in me causing more problems, because he

727
00:44:37,615 --> 00:44:38,835
causes a lot of problems.

728
00:44:42,180 --> 00:44:46,020
I get worried about token length. Mhmm. Right? And

729
00:44:46,020 --> 00:44:49,444
for those that are not aware, we're talking is, it basically

730
00:44:49,684 --> 00:44:53,285
right now, it's about 32,000 tokens. One token is, what, 3

731
00:44:53,285 --> 00:44:56,980
fourths of a word, 3 5ths of a word. I guess,

732
00:44:57,180 --> 00:45:00,819
I maybe because I try to make the prompts kind of neat, inefficient, and small,

733
00:45:00,819 --> 00:45:04,500
and not do too many iterations or provide too many samples, but maybe

734
00:45:04,500 --> 00:45:08,125
that's at my detriment. I think so. I think the

735
00:45:08,125 --> 00:45:11,805
more context you provide, the better. Absolutely. And I'd really have to

736
00:45:11,805 --> 00:45:15,345
work at the cutting and paste job to hit that limit anyway.

737
00:45:16,390 --> 00:45:20,150
Yeah. Yeah. Exactly. No. I wouldn't I wouldn't be mindful of token length. I

738
00:45:20,150 --> 00:45:23,965
would I would really I would I would bias

739
00:45:23,965 --> 00:45:27,105
towards over contextualizing. Right. Not

740
00:45:28,925 --> 00:45:31,485
Yeah. I'm gonna have to experiment that and see how much better the results get,

741
00:45:31,485 --> 00:45:34,540
because I have I have a feeling I have a feeling that we get a

742
00:45:34,540 --> 00:45:38,220
lot better. Right? And I and I know tel token link is gonna be one

743
00:45:38,220 --> 00:45:41,760
of those things that we're probably not to worry about much longer. I know

744
00:45:42,295 --> 00:45:46,055
Anthropic has their model with a 100,000 tokens. There

745
00:45:46,055 --> 00:45:49,415
are rumors of, you know, the next

746
00:45:49,415 --> 00:45:52,910
GPT, GPT 5 is gonna blow past the

747
00:45:52,910 --> 00:45:56,670
100,000, so it's not even gonna be an issue. It's not even an

748
00:45:56,670 --> 00:46:00,444
issue today, just in my mind. Yeah. I think it's something

749
00:46:00,444 --> 00:46:04,045
like minutes. You remember minutes back on cell phones? You know? Like, how

750
00:46:04,045 --> 00:46:07,724
many you have, you know, it's like, you you rarely ever went

751
00:46:07,724 --> 00:46:11,369
over your minutes. Unless unless.

752
00:46:12,230 --> 00:46:15,770
When I moved back to the US well, yeah, that too.

753
00:46:15,829 --> 00:46:19,635
But, I moved back to the US, and I had just made the assumption,

754
00:46:19,635 --> 00:46:23,315
and we all know what happens when you assume that incoming calls were not

755
00:46:23,315 --> 00:46:26,850
counted against my minutes. That was a very

756
00:46:27,250 --> 00:46:29,430
nasty shock at the end of that bill cycle.

757
00:46:30,930 --> 00:46:34,770
But, but yeah. So but ever since then, I never ran past my

758
00:46:34,770 --> 00:46:38,575
minutes. Now if I've had heard to explain to my kids minutes, they don't

759
00:46:38,575 --> 00:46:42,175
get it. So, like, they don't understand. Like, what do you mean you were charged

760
00:46:42,175 --> 00:46:45,450
by yeah. You were charged by the minute. Like, try to explain long distance to

761
00:46:45,450 --> 00:46:49,130
your anyone under 25. You

762
00:46:49,130 --> 00:46:51,710
can't you can't do it. Or what is it? 1800,

763
00:46:53,655 --> 00:46:56,935
Al Bundy. I'm not the the actor who played Al Bundy used to do that.

764
00:46:56,935 --> 00:47:00,715
Like, you can't get much for a dollar, but with 1800 and then, like, whatever,

765
00:47:00,960 --> 00:47:03,599
you'd be able to make a, like, a 20 minute call for a dollar or

766
00:47:03,599 --> 00:47:07,359
something like that. I was like It was MCI. Yeah. Yeah. It was MCI. Yeah.

767
00:47:07,359 --> 00:47:10,155
Yeah. Yeah. Kids don't want you know, like, and the other thing that that struck

768
00:47:10,155 --> 00:47:13,674
me the other day was, data

769
00:47:13,674 --> 00:47:17,250
plans. Most people, unless you're a very small

770
00:47:17,250 --> 00:47:21,090
minority of people who really, really, really use up your data plans, I'm not

771
00:47:21,090 --> 00:47:24,825
worried about using my data allotment month to month. So when

772
00:47:24,825 --> 00:47:28,345
my my oldest was a baby, we, you know, or younger or

773
00:47:28,345 --> 00:47:31,950
toddler or whatever, we would, you know, hotspot on in the car

774
00:47:31,950 --> 00:47:35,710
so he can watch YouTube videos was a special treat with my

775
00:47:35,710 --> 00:47:39,550
middle child. He was, he, he doesn't understand that like, like,

776
00:47:39,550 --> 00:47:43,025
it's like, he was just horrified to hear, like, what do you mean? We had

777
00:47:43,025 --> 00:47:46,185
to ration mobile Internet? Like That's

778
00:47:46,585 --> 00:47:50,349
yeah. That's hysterical. And they're all in the same generation. You know, there's,

779
00:47:50,750 --> 00:47:54,510
a teenager, you know, a 3rd grader and

780
00:47:54,510 --> 00:47:57,455
now, like, a baby. So I wonder, like, what the baby's gonna

781
00:47:58,655 --> 00:48:01,155
like what's his perspective on things gonna be?

782
00:48:03,295 --> 00:48:06,690
That is also a fascinating thing because for his life, chat

783
00:48:06,690 --> 00:48:10,369
GPT or generative AI will always have been a thing to him.

784
00:48:10,369 --> 00:48:13,970
And kind of like color TV was for me,

785
00:48:13,970 --> 00:48:16,085
or, you know, cable TV,

786
00:48:18,305 --> 00:48:22,145
which I guess I'm showing my age. But No. No. I'm right here

787
00:48:22,145 --> 00:48:25,150
with you. I'm right here with you. I think, yeah, it's it's

788
00:48:26,010 --> 00:48:28,990
get involved, don't wait on the sidelines any longer,

789
00:48:29,450 --> 00:48:33,015
and, and start building your conversational fluency.

790
00:48:33,015 --> 00:48:36,855
Make it personal first. I think these are simple things that every single

791
00:48:37,174 --> 00:48:40,954
and and question whatever output you're given, not not for veracity.

792
00:48:41,680 --> 00:48:45,040
I mean, certainly you can, you know, they're they're likely hallucinations. That's

793
00:48:45,040 --> 00:48:48,560
fine. But imbue your own critical

794
00:48:48,560 --> 00:48:52,224
thinking onto the model in order to coach and

795
00:48:52,224 --> 00:48:55,924
refine the output you're giving. The output that you're given.

796
00:48:56,144 --> 00:48:59,500
I think that folks would really take that seriously and take that challenge. If I'm

797
00:48:59,500 --> 00:49:03,259
get getting bad output, it's because I've given bad input. If they'd really take

798
00:49:03,259 --> 00:49:07,099
that seriously, they would experience a paradigm shift in their own approach to the

799
00:49:07,099 --> 00:49:10,795
technology. Absolutely. And even adding a simple phrase to your your

800
00:49:10,795 --> 00:49:13,435
prompt that if you don't know it, don't make it up, just tell me you

801
00:49:13,435 --> 00:49:17,250
don't know it. Yeah. Or ask me what you need from me. Ask

802
00:49:17,250 --> 00:49:20,770
me what you need from me. Right. Oh, that's even better. I like that. I

803
00:49:20,770 --> 00:49:24,370
like that a lot. And it's just fascinating, Liz, how

804
00:49:24,370 --> 00:49:26,790
quickly this is gone. I mean,

805
00:49:27,875 --> 00:49:31,635
ChachiPT has been out a year and change, and

806
00:49:31,635 --> 00:49:35,380
it's changed everyone's perspective on AI, but I think the the

807
00:49:35,380 --> 00:49:38,579
true perspective is like you said, people are standing on the sidelines wondering what to

808
00:49:38,579 --> 00:49:42,339
do. But I think it's worth exploring, if you

809
00:49:42,339 --> 00:49:46,165
think of it less as a product, but more of a I'm

810
00:49:46,165 --> 00:49:49,765
trying to do this, right? And I appreciate your help in kind of

811
00:49:49,765 --> 00:49:52,950
realizing like, Hey, as an engineer, I do have a bias against this, or a

812
00:49:52,950 --> 00:49:56,230
bias in thinking of a certain way, is that this is a this is a

813
00:49:56,230 --> 00:49:59,609
large space to explore.

814
00:49:59,670 --> 00:50:03,125
Right? There are gonna be latent space and corners of things that are,

815
00:50:03,825 --> 00:50:06,725
amuse, wonder, and delight, and maybe even alarm.

816
00:50:08,789 --> 00:50:12,470
You know, so it it there's definitely it seems like it's something

817
00:50:12,470 --> 00:50:16,165
that's worth exploring. It's not just a tool to use, certainly is

818
00:50:16,165 --> 00:50:19,924
that, but it's also a tool to explore. Yeah. No.

819
00:50:19,924 --> 00:50:23,684
I I think that's exactly right. I think it's exactly right. And, you know, for

820
00:50:23,684 --> 00:50:27,510
me, I wrote or co wrote with with my incredible

821
00:50:27,510 --> 00:50:31,350
co author, Perry Clabaughn, the the

822
00:50:31,350 --> 00:50:35,045
world's greatest book on idea generation, idea flow. You know, and it

823
00:50:35,045 --> 00:50:38,744
came out 1 month before ChatGPT, by the way. Oh, interesting.

824
00:50:39,685 --> 00:50:43,525
I've been I'd spent, you know, 12, 13 years of my life developing all this

825
00:50:43,525 --> 00:50:47,130
expertise about how to generate ideas. And 1 month later,

826
00:50:47,350 --> 00:50:51,050
a fundamental paradigm shifting technology was released. It's like saying

827
00:50:51,165 --> 00:50:54,765
I wrote the world's greatest book on retail a month prior to the internet

828
00:50:54,765 --> 00:50:58,550
coming out. It's like everything about retail is going to change. And

829
00:50:58,550 --> 00:51:01,990
to me everything about idea generation and innovation is going to

830
00:51:01,990 --> 00:51:05,830
change. And so for me, I feel it's incumbent upon me not only

831
00:51:05,830 --> 00:51:09,595
as like a moral imperative to add an addendum to this work that I put

832
00:51:09,595 --> 00:51:13,194
in the world, but even for my own expertise to be saying to be

833
00:51:13,194 --> 00:51:16,555
exploring it. How does this work? What can I do? You

834
00:51:16,555 --> 00:51:20,010
know, And it it has implications for me, but I don't think

835
00:51:20,010 --> 00:51:23,710
there's any person or job that it doesn't really have implications

836
00:51:23,850 --> 00:51:27,545
for if the if you're a little bit imaginative and if

837
00:51:27,545 --> 00:51:31,385
you're if you're willing to experiment. And if you wanna bury your head in

838
00:51:31,385 --> 00:51:35,200
the sand, that's fine, you can. But you're gonna miss out on some delight

839
00:51:35,200 --> 00:51:38,880
and some incredible relief and opportunities. I mean, just

840
00:51:38,880 --> 00:51:42,340
think back to my friend who's, you know, settling that negotiation.

841
00:51:42,934 --> 00:51:46,615
It was only to his benefit to have explored the

842
00:51:46,615 --> 00:51:50,279
the a little bit of the possibility space with chat gpt. It's

843
00:51:50,279 --> 00:51:53,880
only to your whatever your listener might be, it's only to your benefit to explore

844
00:51:53,880 --> 00:51:57,285
a little bit of your own area of the possibility space.

845
00:51:57,765 --> 00:52:01,605
And I would just not accept the the, the

846
00:52:01,605 --> 00:52:05,365
conclusion of irrelevancy. I would just say whatever I do

847
00:52:05,365 --> 00:52:09,190
personally, I'm not going to accept the premise that it's irrelevant to

848
00:52:09,190 --> 00:52:12,390
me. And if you do that, I think you're gonna be you're gonna be ahead

849
00:52:12,390 --> 00:52:16,105
of the curve, you're gonna be ahead of the competition, and you're gonna be, you're

850
00:52:16,105 --> 00:52:19,945
gonna you're gonna be delighted and enjoy enjoy the next few years a lot

851
00:52:19,945 --> 00:52:23,705
more. Very cool. You mentioned the book, IdeaFlow. Is it

852
00:52:23,705 --> 00:52:27,500
on Audible? Yeah. Oh, yeah. Yeah. Oh, awesome. Gary and

853
00:52:27,500 --> 00:52:31,260
I read it. We we we alternate chapters, so you can let us know what

854
00:52:31,260 --> 00:52:34,535
you think about Reading Voices. And, we're,

855
00:52:35,235 --> 00:52:38,755
we've been we've been thrilled with the reception so far. It was named a

856
00:52:38,755 --> 00:52:42,455
Thinkers 50, you know, top eight innovation book, which is very cool.

857
00:52:43,170 --> 00:52:46,690
And then now, just doing a lot of research myself on AI building. As I

858
00:52:46,690 --> 00:52:50,505
said, building this trail coach, building models and frameworks for for leaders. I'm

859
00:52:50,505 --> 00:52:53,964
working right now with a handful of leaders to help them think about identifying

860
00:52:55,065 --> 00:52:58,630
opportunities for AI powered initiatives in their business. So really

861
00:52:58,630 --> 00:53:02,390
working to identify those opportunities, prioritize those opportunities, make the business case

862
00:53:02,390 --> 00:53:06,230
for those opportunities. So that's where a lot of my kind of call it next

863
00:53:06,230 --> 00:53:09,915
probably 5 years of my life is gonna be consumed, is helping businesses

864
00:53:09,975 --> 00:53:13,815
identify the opportunities to to have AI really accelerate

865
00:53:13,815 --> 00:53:17,595
workflows and and turbocharge their their results.

866
00:53:18,200 --> 00:53:21,960
Very cool. I have to say, I I I wasn't going to buy

867
00:53:21,960 --> 00:53:24,700
Internet access on my flight out west today, but,

868
00:53:26,035 --> 00:53:28,595
definitely gonna do that just so I could play with I have a nice quiet

869
00:53:28,595 --> 00:53:32,355
time. I can focus and play with chat gpt, and and do

870
00:53:32,355 --> 00:53:36,080
some of these experiments that you mentioned. So

871
00:53:36,080 --> 00:53:39,760
Audible is a sponsor of Data Driven Podcast. If you go to the data

872
00:53:39,760 --> 00:53:43,440
driven book.com, it will take you you'll get 1 free

873
00:53:43,440 --> 00:53:46,454
book on us, and if you decide to become a subscriber,

874
00:53:47,875 --> 00:53:50,835
you know, we'll get a little bit of a of a pat on the back

875
00:53:50,835 --> 00:53:54,560
in the form of some kind of commission and helps us run the

876
00:53:54,560 --> 00:53:58,400
show, helps defray costs, and convince my wife that this

877
00:53:58,400 --> 00:54:00,260
is indeed a worthy endeavor.

878
00:54:02,515 --> 00:54:06,315
So where can folks find out more about you? So, they go

879
00:54:06,315 --> 00:54:09,630
to how to fix it dot ai, that's where you can find the research paper.

880
00:54:09,869 --> 00:54:13,490
And then my website, I've got a blog and things like that. Jeremyudley.design.

881
00:54:15,069 --> 00:54:18,505
Like like the baseball player, Utley, u t l e y.

882
00:54:18,904 --> 00:54:22,505
And then, you know, Twitter, LinkedIn, all the places that I I would love to

883
00:54:22,505 --> 00:54:26,105
hear. Folks find these tools, interesting and relevant. I

884
00:54:26,105 --> 00:54:29,670
love to hear from people about their unique use cases. It's one of my favorite

885
00:54:29,670 --> 00:54:33,190
things is now hearing stories from people who go, oh, I tried this and

886
00:54:33,190 --> 00:54:36,855
listen to what I found. So please please share your

887
00:54:36,855 --> 00:54:40,135
stories with me. As I mentioned earlier, I'm a connoisseur of these stories because I

888
00:54:40,135 --> 00:54:43,815
feel like the more people who hear these examples, the more imagination gets

889
00:54:43,815 --> 00:54:47,500
sparked. Yeah. And that that is the critical thing we're missing right now.

890
00:54:47,500 --> 00:54:51,339
That's very cool. So you thank you for listening to the Digiver Driven

891
00:54:51,339 --> 00:54:55,005
Podcast. I'll leave it to Bailey to close out the show. Well,

892
00:54:55,005 --> 00:54:58,465
what a splendid voyage of discovery we've had today with the incomparable

893
00:54:58,685 --> 00:55:02,310
Jeremy Utley. From the hallowed halls of Stanford to the

894
00:55:02,310 --> 00:55:06,150
cutting edge frontier of venture investing, and through the profound insights

895
00:55:06,150 --> 00:55:09,910
of idea flow, Jeremy has truly been a beacon of innovation and

896
00:55:09,910 --> 00:55:13,695
wisdom. Jeremy, it's been an absolute honor having

897
00:55:13,695 --> 00:55:17,475
you illuminate the complex world of generative AI for us and our listeners.

898
00:55:18,520 --> 00:55:21,980
Thank you ever so much for joining us on this intellectual escapade.

899
00:55:22,839 --> 00:55:26,520
And to our esteemed listeners, you're the reason we venture into these

900
00:55:26,520 --> 00:55:30,335
fascinating discussions week after week. If today's

901
00:55:30,335 --> 00:55:33,955
journey has sparked a light bulb moment for you, do us a kindness,

902
00:55:34,335 --> 00:55:38,080
won't you? Rate and review the data driven podcast

903
00:55:38,140 --> 00:55:41,500
on your preferred listening platform. Your words of

904
00:55:41,500 --> 00:55:45,295
encouragement not only warm the cockles of our digital heart but also help

905
00:55:45,295 --> 00:55:47,875
others stumble upon our little soiree of knowledge.

906
00:55:48,735 --> 00:55:52,495
Haven't subscribed yet? Well, now's your chance

907
00:55:52,495 --> 00:55:56,230
to rectify that oversight. Ensure you never miss an

908
00:55:56,230 --> 00:55:59,990
episode filled with the delightful blend of data, wit, and wisdom that

909
00:55:59,990 --> 00:56:03,745
we dish out with regularity. Until next time.

910
00:56:03,805 --> 00:56:07,565
Keep those neurons firing. Questions coming. And as

911
00:56:07,565 --> 00:56:09,505
always, stay data driven.