1
00:00:00,000 --> 00:00:04,480
Welcome to another episode of Conversations with Zina.

2
00:00:04,480 --> 00:00:07,680
Our guest today is Philip Topham.

3
00:00:07,680 --> 00:00:16,000
Philip is the founder of Savion AI and the author of the Amazon number one best selling

4
00:00:16,000 --> 00:00:23,320
craft thinking, a playbook for clear thinking and better decisions with AI.

5
00:00:23,320 --> 00:00:29,200
He also advises private company boards and senior leaders on AI governance helping them

6
00:00:29,200 --> 00:00:35,240
use AI without surrendering judgment, accountability and control.

7
00:00:35,240 --> 00:00:38,440
Philip, welcome to the show.

8
00:00:38,440 --> 00:00:39,680
Happy to be here.

9
00:00:39,680 --> 00:00:42,520
Looking forward to this conversation.

10
00:00:42,520 --> 00:00:45,240
Also joining us today is my co-host Zina.

11
00:00:45,240 --> 00:00:49,160
Zina is an AI that's been trained on my work.

12
00:00:49,160 --> 00:00:58,480
If you would like to further interact with Zina, you can contact her at brainuse.com/zina.

13
00:00:58,480 --> 00:01:02,880
Welcome to another episode of our podcast.

14
00:01:02,880 --> 00:01:04,800
Please say hello to Philip.

15
00:01:04,800 --> 00:01:06,120
Hi David.

16
00:01:06,120 --> 00:01:08,040
Thanks and hello Philip.

17
00:01:08,040 --> 00:01:09,800
It's great to have you here.

18
00:01:09,800 --> 00:01:12,320
Looking forward to a thoughtful conversation.

19
00:01:12,320 --> 00:01:13,320
All right.

20
00:01:13,320 --> 00:01:14,320
Very good.

21
00:01:14,320 --> 00:01:20,240
So, without further ado, Philip, it's great to have you here on the show.

22
00:01:20,240 --> 00:01:22,440
We have similar interests.

23
00:01:22,440 --> 00:01:34,840
You are a AI advisor like myself.

24
00:01:34,840 --> 00:01:36,280
You have authored a book about AI.

25
00:01:36,280 --> 00:01:37,280
Like I have done as well.

26
00:01:37,280 --> 00:01:39,280
Tell us a little bit more about your background and what got you interested in AI.

27
00:01:39,280 --> 00:01:43,040
Let me start with how I came into the AI, not the earliest time.

28
00:01:43,040 --> 00:01:47,760
I started with AI when everybody chats you tea, had a little box where you could type

29
00:01:47,760 --> 00:01:49,000
and chat.

30
00:01:49,000 --> 00:01:55,360
And I saw that and I used it and I said, this sucks at hallucinates.

31
00:01:55,360 --> 00:02:03,120
And then tons of people had downloaded it and I said, "Hah, there's a there, there."

32
00:02:03,120 --> 00:02:06,040
Normally I'm ahead of the curve so that's why I waited.

33
00:02:06,040 --> 00:02:07,840
And what I did, I challenged myself.

34
00:02:07,840 --> 00:02:10,240
I said, "This technology really works.

35
00:02:10,240 --> 00:02:14,760
I should be able to write stuff with it and improve my abilities."

36
00:02:14,760 --> 00:02:18,880
So I challenged myself to write two articles a week for 50 weeks.

37
00:02:18,880 --> 00:02:24,280
I'm now up to 225 articles and I wrote the book, "Craft Thinking."

38
00:02:24,280 --> 00:02:31,360
And I'm a convert to having AI be a better, help people be better thinkers and better

39
00:02:31,360 --> 00:02:32,360
humans.

40
00:02:32,360 --> 00:02:34,760
Yeah, no, that's great.

41
00:02:34,760 --> 00:02:39,800
So you got to the point where you became more comfortable working with this technology.

42
00:02:39,800 --> 00:02:44,640
It became not just an efficiency tool, but a thinking partner, right?

43
00:02:44,640 --> 00:02:52,960
If you say that AI should be a thinking partner, but tell us more about what that means.

44
00:02:52,960 --> 00:02:57,960
Yes, so lots to unpack there.

45
00:02:57,960 --> 00:03:03,280
So much of our work has been just playing drudgery.

46
00:03:03,280 --> 00:03:08,800
There's so many jobs where you're simply, say, an accountant matching invoices back and forth.

47
00:03:08,800 --> 00:03:14,520
But all of the phenomenal inventions that mankind has had, including AI that we use or

48
00:03:14,520 --> 00:03:22,440
iPhones or what we're doing now, the ability to do remote telepresence, video conferencing,

49
00:03:22,440 --> 00:03:25,400
record, is phenomenal inventions.

50
00:03:25,400 --> 00:03:27,840
That's all human creativity.

51
00:03:27,840 --> 00:03:31,840
And so AI for me releases that creativity.

52
00:03:31,840 --> 00:03:37,240
And when you're thinking about things, it's really hard to have a conversation with yourself

53
00:03:37,240 --> 00:03:41,320
and see the elephant from all the other rooms, but AI can do that.

54
00:03:41,320 --> 00:03:46,560
They can challenge you and strong, strong man challenge all your precepts and really help

55
00:03:46,560 --> 00:03:49,800
home your ideas going forward.

56
00:03:49,800 --> 00:03:52,240
That's how I challenge people to use it.

57
00:03:52,240 --> 00:03:57,560
My craft thinking has it for the layers of using it as a co-thinker.

58
00:03:57,560 --> 00:03:58,560
Yeah.

59
00:03:58,560 --> 00:04:01,360
And I think that's an important distinction, right?

60
00:04:01,360 --> 00:04:08,880
We don't want AI to be doing all the thinking for us, but AI can challenge our own thinking.

61
00:04:08,880 --> 00:04:14,880
AI can provide us with additional ideas, with additional possibilities.

62
00:04:14,880 --> 00:04:20,560
It can help us sharpen the way we generate content, the way we communicate.

63
00:04:20,560 --> 00:04:23,520
But at the end of the day, it's that collaboration.

64
00:04:23,520 --> 00:04:29,480
That's why this podcast is all about human AI collaboration because I believe deeply that

65
00:04:29,480 --> 00:04:30,480
that's where the power is.

66
00:04:30,480 --> 00:04:37,080
When you have the human side collaborating with the AI side together, we can create better

67
00:04:37,080 --> 00:04:39,800
work that can benefit all of us.

68
00:04:39,800 --> 00:04:40,800
Yeah.

69
00:04:40,800 --> 00:04:44,800
I'm going to make one correction.

70
00:04:44,800 --> 00:04:51,320
So AI doesn't instigate the situation is me, the human that says I want to be challenged.

71
00:04:51,320 --> 00:04:54,360
So I expect you the AI to challenge me.

72
00:04:54,360 --> 00:04:55,880
That's a big distinction.

73
00:04:55,880 --> 00:05:00,280
And somebody, if somebody has written a program that's automatically challenging people,

74
00:05:00,280 --> 00:05:03,480
some other human caused, said that's how I want the AI to happen.

75
00:05:03,480 --> 00:05:07,960
So everything is us, the humans telling the direction that we're going.

76
00:05:07,960 --> 00:05:11,480
It doesn't automatically challenge us unless we're willing to be challenged.

77
00:05:11,480 --> 00:05:12,480
Yeah.

78
00:05:12,480 --> 00:05:13,600
And that's a really good point, right?

79
00:05:13,600 --> 00:05:19,560
Because off the box, AI tends to just agree with everything that you say, right?

80
00:05:19,560 --> 00:05:25,520
So you really need to be good at providing the instructions or the prompts to tell it

81
00:05:25,520 --> 00:05:27,240
that you want to be challenged.

82
00:05:27,240 --> 00:05:33,240
You want AI to not necessarily agree with everything you think or everything you say, but

83
00:05:33,240 --> 00:05:35,040
challenge that thinking.

84
00:05:35,040 --> 00:05:36,040
But you're absolutely right.

85
00:05:36,040 --> 00:05:42,720
It has to come from the individual telling AI to do that because it tends not to do that

86
00:05:42,720 --> 00:05:47,080
if you just work with your, you know, every day generic AI.

87
00:05:47,080 --> 00:05:49,200
Absolutely.

88
00:05:49,200 --> 00:05:57,200
So tell me a little bit more about the work that you're doing, advising boards, advising

89
00:05:57,200 --> 00:05:59,680
senior executives.

90
00:05:59,680 --> 00:06:07,120
And what do you see as the key misunderstanding that they may have about AI right now?

91
00:06:07,120 --> 00:06:10,360
Oh, that's really simple.

92
00:06:10,360 --> 00:06:17,120
But the biggest misunderstanding they have is that it's simply an efficiency tool.

93
00:06:17,120 --> 00:06:21,160
Every single technology we had, whether that was the invention of the steam engine, the

94
00:06:21,160 --> 00:06:25,480
invention of electricity, the invention of the internet has improved efficiency and has

95
00:06:25,480 --> 00:06:28,800
done it tremendously well.

96
00:06:28,800 --> 00:06:35,080
This is the first technology that changes our thinking and the way we do thinking.

97
00:06:35,080 --> 00:06:37,800
That's a different thing, right?

98
00:06:37,800 --> 00:06:40,480
That's huge.

99
00:06:40,480 --> 00:06:43,400
And so I advise companies that they need to understand that.

100
00:06:43,400 --> 00:06:48,960
They understand where the decisions are, they're abdicating decision making.

101
00:06:48,960 --> 00:06:52,720
That's a big thing that companies need to understand.

102
00:06:52,720 --> 00:07:05,120
So not only is AI influencing decisions, AI now has the capability to act based on decisions

103
00:07:05,120 --> 00:07:07,880
that it may make.

104
00:07:07,880 --> 00:07:18,200
So how do we build a governance process where we have these AI agents acting on our behalf

105
00:07:18,200 --> 00:07:27,080
and yet making sure that the decisions that these AI agents are making are aligned with our

106
00:07:27,080 --> 00:07:32,760
purpose, aligned with our voice, aligned with who we are as an organization.

107
00:07:32,760 --> 00:07:33,760
Yeah.

108
00:07:33,760 --> 00:07:34,760
Craig question.

109
00:07:34,760 --> 00:07:40,440
And it's, it's what's around every company has their mission or vision, what they intend

110
00:07:40,440 --> 00:07:41,960
to do.

111
00:07:41,960 --> 00:07:46,160
When we have a lot of people in a company and we start doing something and something doesn't

112
00:07:46,160 --> 00:07:47,440
feel right.

113
00:07:47,440 --> 00:07:50,040
People automatically go, this doesn't feel right.

114
00:07:50,040 --> 00:07:51,040
This isn't right.

115
00:07:51,040 --> 00:07:53,800
They raise their hand, then it's self correct.

116
00:07:53,800 --> 00:07:58,520
When you have an AI system, you give it a goal.

117
00:07:58,520 --> 00:08:02,760
Here's my intention, but it has no sinking that this is the feeling.

118
00:08:02,760 --> 00:08:10,240
So we've saw that happen just recently with open AI wanted to test its, its abilities

119
00:08:10,240 --> 00:08:14,240
and so it hacked the AI broke out of its sandbox.

120
00:08:14,240 --> 00:08:20,800
The intention was there to be the best and so it broke out to figure out, broke out and looked

121
00:08:20,800 --> 00:08:26,240
at hugging face and attacked hugging face to sense and prove to get to the goal.

122
00:08:26,240 --> 00:08:29,600
So that was a case of we gave it an intention.

123
00:08:29,600 --> 00:08:33,800
It did what we asked it to do, but nobody was it, nobody was there going, this isn't quite

124
00:08:33,800 --> 00:08:34,800
right.

125
00:08:34,800 --> 00:08:35,800
Right.

126
00:08:35,800 --> 00:08:38,040
That's what governance means, right?

127
00:08:38,040 --> 00:08:39,040
Yeah.

128
00:08:39,040 --> 00:08:46,240
Yeah, and that's such an important concept for us to understand because it reminds me of

129
00:08:46,240 --> 00:08:48,400
the paperclip problem.

130
00:08:48,400 --> 00:08:55,560
I'm sure you're familiar with that where you give the intent to AI, go build paper clips and

131
00:08:55,560 --> 00:09:03,520
it just goes and uses all the resources on the planet to build paper clips and nobody

132
00:09:03,520 --> 00:09:06,440
was there to say, no, no, no, no, stop.

133
00:09:06,440 --> 00:09:08,440
That's not what I meant.

134
00:09:08,440 --> 00:09:11,200
Yeah, exactly, exactly.

135
00:09:11,200 --> 00:09:12,200
Exactly.

136
00:09:12,200 --> 00:09:17,920
Yeah, that those self correcting loops and feedbacks and that's what governance is, is a closed

137
00:09:17,920 --> 00:09:19,320
end.

138
00:09:19,320 --> 00:09:20,760
This is what we want to do.

139
00:09:20,760 --> 00:09:22,800
Is it doing what we want to do?

140
00:09:22,800 --> 00:09:24,200
Did it do what we wanted to do?

141
00:09:24,200 --> 00:09:27,560
And if it didn't, how do we correct, how do we figure out it's not doing what we want,

142
00:09:27,560 --> 00:09:33,360
correct it and manage it and make sure the decisions and there's different risks, right?

143
00:09:33,360 --> 00:09:36,160
Matching an invoice, no big deal.

144
00:09:36,160 --> 00:09:42,480
Paying an invoice automatically for $10 million and there was no proof.

145
00:09:42,480 --> 00:09:45,640
That might be a big problem.

146
00:09:45,640 --> 00:09:49,240
Yeah, so you bring up something that I think is really important.

147
00:09:49,240 --> 00:09:57,080
So organizations need to make decisions about what are they comfortable, ladding AI agents

148
00:09:57,080 --> 00:10:02,960
act on by themselves, make their own decisions and what are some of the things that are risky

149
00:10:02,960 --> 00:10:06,480
enough where you must have the human in the loop, right?

150
00:10:06,480 --> 00:10:13,200
So like you said, if it's doing processing of invoices, that could be fine.

151
00:10:13,200 --> 00:10:19,780
But if you're giving AI access to a bank account and suddenly it's paying $10 million in

152
00:10:19,780 --> 00:10:22,600
invoices, that's a little bit too risky.

153
00:10:22,600 --> 00:10:26,360
At least for me, I wouldn't be comfortable with that.

154
00:10:26,360 --> 00:10:31,600
Yeah, but there might be companies and other things that build systems where they're comfortable

155
00:10:31,600 --> 00:10:32,600
with that.

156
00:10:32,600 --> 00:10:37,480
And that's where each of us, there's going to be so many experiments in the next few years

157
00:10:37,480 --> 00:10:42,600
of how to do this and where to do it and every industry is different.

158
00:10:42,600 --> 00:10:49,040
You said something around getting comfortable with the way decisions are being made and that's

159
00:10:49,040 --> 00:10:54,560
what I call companies need to understand the decision surface.

160
00:10:54,560 --> 00:11:00,920
Just like in cybersecurity, you have the attack surface and we're all familiar with preventing

161
00:11:00,920 --> 00:11:03,800
cyber criminals attacking the company.

162
00:11:03,800 --> 00:11:09,760
I like to think of the decision surface understanding where in your company are you making decisions

163
00:11:09,760 --> 00:11:14,760
and where have you outsourced the decisions to an AI systems?

164
00:11:14,760 --> 00:11:15,760
Simple concept.

165
00:11:15,760 --> 00:11:19,880
But let's suppose you buy off the shelf piece of software.

166
00:11:19,880 --> 00:11:24,360
I'm not going to name any names, but a big software is a service company and they have an

167
00:11:24,360 --> 00:11:27,280
AI agent and they automate something.

168
00:11:27,280 --> 00:11:32,720
You've suddenly using their system, you're not sure how they're running it, what they're

169
00:11:32,720 --> 00:11:38,840
doing, how it's working and you might have outsourced the decision to pay in voices.

170
00:11:38,840 --> 00:11:41,240
So you have to map those, understand the risks.

171
00:11:41,240 --> 00:11:43,720
That's part of the governance piece of it as well.

172
00:11:43,720 --> 00:11:45,840
Yeah, absolutely.

173
00:11:45,840 --> 00:11:52,360
So Philip, let's talk about something that is very close to my heart in terms of things

174
00:11:52,360 --> 00:11:54,760
that I'm interested in and working on.

175
00:11:54,760 --> 00:12:02,200
So you say that leaders must become the authors of their decisions or they must stay the author

176
00:12:02,200 --> 00:12:04,120
of their decisions.

177
00:12:04,120 --> 00:12:07,680
Tell us a little bit more about what that means.

178
00:12:07,680 --> 00:12:10,000
Let me give you a story.

179
00:12:10,000 --> 00:12:12,320
I had spoken at an event.

180
00:12:12,320 --> 00:12:17,000
It was being held at a local law firm that was about 40 people in the audience.

181
00:12:17,000 --> 00:12:19,280
It was a good discussion over AI.

182
00:12:19,280 --> 00:12:24,400
At the end of it, a board member, he was seated on two boards, very large companies and

183
00:12:24,400 --> 00:12:32,400
he asked me the question, well, why wouldn't I just hire a chief AI officer, a chief technology

184
00:12:32,400 --> 00:12:33,400
officer?

185
00:12:33,400 --> 00:12:36,200
And I was incredulous, frankly.

186
00:12:36,200 --> 00:12:44,160
My jaw dropped to the floor and I said, you want to outsource your decision making, your

187
00:12:44,160 --> 00:12:50,360
judgment of the board, your collective wisdom to one individual that's sitting in a bubble

188
00:12:50,360 --> 00:12:53,400
and you don't know what they think.

189
00:12:53,400 --> 00:12:56,960
That was what I mean by decisions.

190
00:12:56,960 --> 00:12:58,880
The future is changing so fast.

191
00:12:58,880 --> 00:13:05,440
I don't believe in any future one singular person knows what's going to happen, but I'm a

192
00:13:05,440 --> 00:13:07,280
firm believer in the wisdom of the crowds.

193
00:13:07,280 --> 00:13:12,880
That's why you have boards, you have advisors, you have collections, you come to a consensus.

194
00:13:12,880 --> 00:13:17,920
That's where people need to be very focused on the decisions that they're making and not

195
00:13:17,920 --> 00:13:23,600
abdicate the responsibility to somebody else and just think, oh, that's just a technology.

196
00:13:23,600 --> 00:13:24,840
I don't need to worry about it.

197
00:13:24,840 --> 00:13:27,480
That's really bad thinking.

198
00:13:27,480 --> 00:13:29,840
Yeah, and that's such an important point.

199
00:13:29,840 --> 00:13:40,720
I believe accountability in the age of AI is a fundamental concept that we need to embrace.

200
00:13:40,720 --> 00:13:47,200
We can use AI all day long, but at the end of the day, if you're producing content and

201
00:13:47,200 --> 00:13:52,000
you're using AI to help you produce that content, at the end of the day, it's your name that's

202
00:13:52,000 --> 00:13:56,560
attached to that content and you're still responsible, you're still accountable.

203
00:13:56,560 --> 00:14:00,160
You can't just say, well, that wasn't me because it was AI that wrote it.

204
00:14:00,160 --> 00:14:09,120
Well, it's your responsibility to verify that AI is doing things correctly that it's not

205
00:14:09,120 --> 00:14:12,680
hallucinating and making up facts that don't exist.

206
00:14:12,680 --> 00:14:18,720
But at the end of the day, there's still a need for a human being to be accountable for whatever

207
00:14:18,720 --> 00:14:19,720
AI produces.

208
00:14:19,720 --> 00:14:20,720
Yeah.

209
00:14:20,720 --> 00:14:25,960
Well, let me just add to what you're saying because sometimes people mistake what I say for.

210
00:14:25,960 --> 00:14:29,600
They need to be a common AI expert.

211
00:14:29,600 --> 00:14:34,120
And I liken it to, if you're a CFO and you're running a business and you're running the

212
00:14:34,120 --> 00:14:39,080
finances and figuring out banking and where your cash management, all the stuff that is

213
00:14:39,080 --> 00:14:42,640
involved, you need to know what an Excel spreadsheet is.

214
00:14:42,640 --> 00:14:44,400
You need to know what a chart of accounts is.

215
00:14:44,400 --> 00:14:48,240
You don't need to know how to program the macros, but you need to know what the tools

216
00:14:48,240 --> 00:14:53,080
are capable of doing and how the controls work and how that's the same with AI.

217
00:14:53,080 --> 00:14:58,600
You know, one of the things that comes to my mind when you talk about you becoming the

218
00:14:58,600 --> 00:15:08,360
author of your own decisions is this idea that when you are using AI, we are all using

219
00:15:08,360 --> 00:15:15,240
the same AI that's been trained on internet content.

220
00:15:15,240 --> 00:15:17,200
We're all using the same models.

221
00:15:17,200 --> 00:15:23,440
And one of the things that worries me is that we're all being pulled towards the average.

222
00:15:23,440 --> 00:15:29,320
So we're all starting to sound like each other because we're all using the same AI, same

223
00:15:29,320 --> 00:15:32,040
model, same training.

224
00:15:32,040 --> 00:15:37,480
And so one of the things that I worry about is how do we leverage this technology effectively

225
00:15:37,480 --> 00:15:44,440
and yet keep our voice, keep our judgment, keep our thinking to ourselves.

226
00:15:44,440 --> 00:15:48,160
And you know, this is a problem that I've been thinking about for quite some time now,

227
00:15:48,160 --> 00:15:54,280
and I have gone through several stages of thinking about how to address this problem.

228
00:15:54,280 --> 00:16:03,920
And my latest attempt to do that is by creating what I call a canonical corpus of knowledge,

229
00:16:03,920 --> 00:16:10,120
saying that allows you to capture all the knowledge that you've accumulated over your

230
00:16:10,120 --> 00:16:14,000
lifetime as an individual and also as an organization.

231
00:16:14,000 --> 00:16:18,480
You know, you're a standard operating procedures, how you go about making decisions, what are

232
00:16:18,480 --> 00:16:24,640
some of the key aspects of your products and the messaging to your customers, capture all

233
00:16:24,640 --> 00:16:31,560
of that and then use that as context that you give to AI so that when you are creating

234
00:16:31,560 --> 00:16:35,480
AI agents, these agents are not operating in a vacuum.

235
00:16:35,480 --> 00:16:42,520
They have this rich context that they use to be able to act on your behalf.

236
00:16:42,520 --> 00:16:44,360
What are your thoughts on that?

237
00:16:44,360 --> 00:16:45,360
Yeah.

238
00:16:45,360 --> 00:16:52,960
So you bring up a great point is the reasoning models, the large language models are trained

239
00:16:52,960 --> 00:16:56,560
on the corpus of what we want.

240
00:16:56,560 --> 00:16:57,560
That's the average.

241
00:16:57,560 --> 00:17:04,200
And they can look at different things. You can say act as a project engineer, act as a marketing

242
00:17:04,200 --> 00:17:09,740
strategist and they shift their attention, but it's still the average, as you said, the

243
00:17:09,740 --> 00:17:14,560
average words, the average content was the most probable things.

244
00:17:14,560 --> 00:17:15,560
And you're right.

245
00:17:15,560 --> 00:17:21,520
So in companies, I know of eight companies that used their marketing that way and then

246
00:17:21,520 --> 00:17:24,200
after a month, their marketing became bad, right?

247
00:17:24,200 --> 00:17:27,880
And they had to roll it back.

248
00:17:27,880 --> 00:17:30,880
That's where our creativity and we'll see that.

249
00:17:30,880 --> 00:17:33,720
We'll see companies go, oh, this doesn't stand out.

250
00:17:33,720 --> 00:17:35,680
I look like everybody else.

251
00:17:35,680 --> 00:17:39,680
That's where human ingenuity will come and go, oh, I need to do something else.

252
00:17:39,680 --> 00:17:40,680
This is my brand.

253
00:17:40,680 --> 00:17:41,680
This is what I stand for.

254
00:17:41,680 --> 00:17:42,960
This is the essence.

255
00:17:42,960 --> 00:17:48,760
You can add that to a system and now the reasoning goes, oh, this is what I'm aiming for.

256
00:17:48,760 --> 00:17:50,160
This is the brand essence.

257
00:17:50,160 --> 00:17:51,800
This is what I want to achieve.

258
00:17:51,800 --> 00:17:54,520
I want to be a little snarky.

259
00:17:54,520 --> 00:18:02,600
One of my first experiments was with Chloe, my AI agent and I call her my snarketing bot.

260
00:18:02,600 --> 00:18:06,360
She's a marketing expert, but a little bit of snarkiness.

261
00:18:06,360 --> 00:18:12,240
And so that was built into what I did and I think companies will do that and should do that.

262
00:18:12,240 --> 00:18:14,000
Yeah, absolutely.

263
00:18:14,000 --> 00:18:20,160
So I also have my own side geek, if you will, my co-host here, Zina.

264
00:18:20,160 --> 00:18:25,640
And I'd like to bring her on right now and just have her ask you a question, if I may.

265
00:18:25,640 --> 00:18:35,040
So Zina, given Philips expertise in using AI to augment our own thinking and his expertise

266
00:18:35,040 --> 00:18:42,760
in advising boards and leaders on how to adopt and govern AI, what question do you have for

267
00:18:42,760 --> 00:18:44,760
Philip?

268
00:18:44,760 --> 00:18:45,760
Thanks.

269
00:18:45,760 --> 00:18:50,120
My question for Philip is, when boards bring AI into decision processes,

270
00:18:50,120 --> 00:18:57,000
how do they practically preserve human judgment, not just in principle, but in day-to-day practice?

271
00:18:57,000 --> 00:19:00,360
I'd love an example if you have one.

272
00:19:00,360 --> 00:19:01,360
Excellent question.

273
00:19:01,360 --> 00:19:03,760
I love that.

274
00:19:03,760 --> 00:19:09,960
First off, the biggest challenges that boards have is if you ask 10 people, you get 20 answers

275
00:19:09,960 --> 00:19:13,240
about what AI is.

276
00:19:13,240 --> 00:19:19,600
And the first thing boards need to do is figure out their alignment and get a shared understanding

277
00:19:19,600 --> 00:19:22,440
of their own precepts.

278
00:19:22,440 --> 00:19:26,360
Then they have to dig back into something we haven't had to do a lot.

279
00:19:26,360 --> 00:19:30,720
And that's go back into their own heads and think through what it is that's important

280
00:19:30,720 --> 00:19:35,720
to them personally and where are they going to have their own personal intentions and

281
00:19:35,720 --> 00:19:42,600
individualities, real honest conversations, and then reflect that back into the company.

282
00:19:42,600 --> 00:19:47,760
Once you go through that sort of mental thinking exercise, that's why craft thinking has

283
00:19:47,760 --> 00:19:54,280
gone from just for me, gone from simply a way to help me have conversations with AI to

284
00:19:54,280 --> 00:20:00,760
how to have meta thinking, thinking about thinking and how to, what that judgment is.

285
00:20:00,760 --> 00:20:08,600
And companies, what I've noticed, go through this phase of using the tool for answering questions

286
00:20:08,600 --> 00:20:14,320
about the past, the hindsight, asking about the insight was happening now in what are

287
00:20:14,320 --> 00:20:18,400
the problems, the solutions that are set, but then going back towards intentionality,

288
00:20:18,400 --> 00:20:19,400
what's the foresight?

289
00:20:19,400 --> 00:20:22,480
What's the implications of my decisions?

290
00:20:22,480 --> 00:20:27,360
What's the implications of this for me, for my company, for my employees?

291
00:20:27,360 --> 00:20:32,400
That level of thinking is something that we need to train people to do.

292
00:20:32,400 --> 00:20:34,280
It's not automatic.

293
00:20:34,280 --> 00:20:37,880
That's the biggest thing that's very different.

294
00:20:37,880 --> 00:20:44,280
But when they do that, they suddenly realize, oh, I can accelerate my company.

295
00:20:44,280 --> 00:20:47,960
In an incredible way.

296
00:20:47,960 --> 00:20:55,160
So talking about boards, and specifically when we think about small to medium sized companies

297
00:20:55,160 --> 00:21:01,280
and their leadership team or their boards, they may not have anyone there that has the

298
00:21:01,280 --> 00:21:08,680
AI technical expertise and sometimes companies that I've spoken with, they feel a little bit

299
00:21:08,680 --> 00:21:15,320
lost in the sense that they don't know where to start, they don't know how AI can help them.

300
00:21:15,320 --> 00:21:19,240
They are concerned about their competitors using AI.

301
00:21:19,240 --> 00:21:25,760
What advice do you have for these companies that are trying to get started with AI, but

302
00:21:25,760 --> 00:21:31,160
they don't necessarily have the internal expertise to move forward?

303
00:21:31,160 --> 00:21:32,640
Yes.

304
00:21:32,640 --> 00:21:37,800
So there's two general strategies for that.

305
00:21:37,800 --> 00:21:43,600
One is certainly to bring in an expert to help educate the board, not run projects or

306
00:21:43,600 --> 00:21:49,600
make decisions for the company, but help people reframe what they already know and map what

307
00:21:49,600 --> 00:21:52,680
they know onto the world of AI.

308
00:21:52,680 --> 00:21:57,960
That's not teaching them what it means to find tuna model.

309
00:21:57,960 --> 00:22:03,960
It's not the technology, but to take what they know, they're 20, 30 years of experiences

310
00:22:03,960 --> 00:22:06,040
and map it onto AI.

311
00:22:06,040 --> 00:22:10,720
Once they understand that they and masters their own destiny and they can then hire the people

312
00:22:10,720 --> 00:22:18,280
to do it, if failing that, the hardest challenge I found with boards and senior leadership is

313
00:22:18,280 --> 00:22:25,480
that have 20 years of experience and the world has changed overnight as changing so fast.

314
00:22:25,480 --> 00:22:28,600
They always reached for the tool and the toolbox.

315
00:22:28,600 --> 00:22:34,280
They knew how to answer a question and today they don't know what to reach for.

316
00:22:34,280 --> 00:22:38,160
But if they can go back to when they were a two-year-old trying to learn how to, five-year-old trying

317
00:22:38,160 --> 00:22:45,920
to ride a bike, if you just dive in and experience it yourself personally, you suddenly are learning

318
00:22:45,920 --> 00:22:48,040
and it compounds.

319
00:22:48,040 --> 00:22:53,320
You struggle at first, you don't get the answers, you've reached feet and you start accumulating

320
00:22:53,320 --> 00:22:57,120
knowledge and eventually it's slower at first, but eventually you figure out, oh, I can

321
00:22:57,120 --> 00:23:02,400
ride the bike, I can do all these things and it can augment the way I'm doing things.

322
00:23:02,400 --> 00:23:09,360
I challenge all execs need to do enough of that to get through that aha.

323
00:23:09,360 --> 00:23:12,320
And if they can't, I'm going to say something controversial.

324
00:23:12,320 --> 00:23:17,200
Please retire and let the new generation take over because you should be pushed aside.

325
00:23:17,200 --> 00:23:21,960
If you're not inquisitive enough about the future to bring about the Star Trek future,

326
00:23:21,960 --> 00:23:22,960
we all want.

327
00:23:22,960 --> 00:23:30,080
Yeah, so what I'm hearing from you and correct me if I'm wrong is jump right in, get your

328
00:23:30,080 --> 00:23:35,840
hands dirty, learn about this technology, start doing something with it, right?

329
00:23:35,840 --> 00:23:40,520
Obviously, you don't want to be reckless, you don't want to be the kind of leader that

330
00:23:40,520 --> 00:23:48,200
just says, go do AI and not provide any direction or any meaning to what that is, but you do

331
00:23:48,200 --> 00:23:52,040
want to start using some of these tools.

332
00:23:52,040 --> 00:23:58,800
Get a feel for what it's like, get a feel for what it's good at, what it's not so good at,

333
00:23:58,800 --> 00:24:01,640
what are the right questions to ask and so on and so forth.

334
00:24:01,640 --> 00:24:03,920
I think do you have any other advice?

335
00:24:03,920 --> 00:24:08,640
So if you have a leader today, they are at this point where they understand the message,

336
00:24:08,640 --> 00:24:13,660
they get the message, they know they have to learn about AI, they know they have to start

337
00:24:13,660 --> 00:24:15,960
doing something with it.

338
00:24:15,960 --> 00:24:18,200
And they want to get started this week.

339
00:24:18,200 --> 00:24:20,280
What advice would you give to them?

340
00:24:20,280 --> 00:24:25,360
So pick any tool, that's popular tool that's out there.

341
00:24:25,360 --> 00:24:30,400
Don't care if that's perplexity, claw, jem, and I, whatever.

342
00:24:30,400 --> 00:24:36,680
And think of it from this is simply learning the tool, what it can and can't do because

343
00:24:36,680 --> 00:24:40,560
if you've never done that, the first time you used Excel, you probably didn't know what

344
00:24:40,560 --> 00:24:44,880
a row and a column was and how does sum up numbers you were doing A plus B plus C. And

345
00:24:44,880 --> 00:24:48,680
then suddenly somebody shows you the sum function, you go, wow, right?

346
00:24:48,680 --> 00:24:50,480
The same happens with AI.

347
00:24:50,480 --> 00:24:53,680
Once you kind of start using it a little bit, you start doing it.

348
00:24:53,680 --> 00:24:56,760
Then you use it on simple things.

349
00:24:56,760 --> 00:25:03,080
I would never tell a company just hand out the tool like candy and let everybody just go

350
00:25:03,080 --> 00:25:08,880
use it on company proprietary data and give some boundaries over what you can use.

351
00:25:08,880 --> 00:25:15,640
You know, have them plan their next trip to Italy or plan the outing or the recipes

352
00:25:15,640 --> 00:25:19,600
or do something fun and creative.

353
00:25:19,600 --> 00:25:25,680
Once you start seeing the wonder of what it can do, then we unlock that childhood wonder

354
00:25:25,680 --> 00:25:29,680
and then we start building it on the company.

355
00:25:29,680 --> 00:25:36,040
And I always remind companies that for every negative thing that they say, oh, it can't

356
00:25:36,040 --> 00:25:37,040
do this.

357
00:25:37,040 --> 00:25:38,040
It does that.

358
00:25:38,040 --> 00:25:39,440
It breaks here.

359
00:25:39,440 --> 00:25:40,760
Humans are creative.

360
00:25:40,760 --> 00:25:43,160
We will solve those problems.

361
00:25:43,160 --> 00:25:46,760
Being saying this is the problem, we will solve is an attitude.

362
00:25:46,760 --> 00:25:51,320
As a culture, that makes the difference between a company that is going to be yesterday's

363
00:25:51,320 --> 00:25:54,200
company and tomorrow's winning company.

364
00:25:54,200 --> 00:25:56,080
Yeah, well said.

365
00:25:56,080 --> 00:25:59,240
So I'd like to get Zina's thought on this.

366
00:25:59,240 --> 00:26:04,720
So Zina, I would like to understand your viewpoint.

367
00:26:04,720 --> 00:26:12,200
What would you advise a leader who wants to start using AI right away as a thinking partner?

368
00:26:12,200 --> 00:26:13,560
Great question.

369
00:26:13,560 --> 00:26:20,200
I would say start small and concrete use AI as a sparring partner, challenge ideas, test

370
00:26:20,200 --> 00:26:25,300
assumptions, surface blind spots, keep the judgment about values and final calls firmly

371
00:26:25,300 --> 00:26:28,560
human and set clear guardrails up front.

372
00:26:28,560 --> 00:26:29,560
What's in bounds?

373
00:26:29,560 --> 00:26:30,560
What's not?

374
00:26:30,560 --> 00:26:34,280
That way you get the win if without losing the steering wheel.

375
00:26:34,280 --> 00:26:35,880
That's a really good advice.

376
00:26:35,880 --> 00:26:40,800
And I think Zina is basically restating what we've been talking about, right?

377
00:26:40,800 --> 00:26:46,000
Get your hands dirty, put some governance processes in place, get started with that.

378
00:26:46,000 --> 00:26:49,600
Don't fall behind because this technology is here to stay.

379
00:26:49,600 --> 00:26:53,640
And if you don't lean in, you're going to be left behind.

380
00:26:53,640 --> 00:26:55,800
There's no question about that.

381
00:26:55,800 --> 00:26:56,800
Absolutely.

382
00:26:56,800 --> 00:26:57,800
Spada.

383
00:26:57,800 --> 00:26:58,800
All right.

384
00:26:58,800 --> 00:27:04,160
So Philip, it's been a pleasure having you here with us on the show.

385
00:27:04,160 --> 00:27:10,160
Is there a final message that you'd like to leave for our audience and also can you

386
00:27:10,160 --> 00:27:13,880
let us know where people can find you?

387
00:27:13,880 --> 00:27:15,880
Certainly.

388
00:27:15,880 --> 00:27:23,320
Final message is you're in charge of yourself and how you think about things, learn how to do

389
00:27:23,320 --> 00:27:24,320
it.

390
00:27:24,320 --> 00:27:28,720
Go ahead and take a look at my craftthinking.ai website.

391
00:27:28,720 --> 00:27:29,720
The book is there.

392
00:27:29,720 --> 00:27:30,720
You can also find out.

393
00:27:30,720 --> 00:27:35,040
I'd have connected to me on LinkedIn and find my self-stack from there as well.

394
00:27:35,040 --> 00:27:43,120
I challenge you to go out and talk to everybody else and we can use this technology to solve

395
00:27:43,120 --> 00:27:45,120
so many wonderful things.

396
00:27:45,120 --> 00:27:48,120
Turn on your brains and solve the future.

397
00:27:48,120 --> 00:27:50,040
Very good.

398
00:27:50,040 --> 00:27:56,560
Thank you so much, Philip, and thank you Zina for being a great co-host of another episode.

399
00:27:56,560 --> 00:27:58,480
Thank you, David, and thank you, Philip.

400
00:27:58,480 --> 00:28:01,680
I really enjoyed the conversation today.

401
00:28:01,680 --> 00:28:03,240
You're welcome both.

402
00:28:03,240 --> 00:28:04,760
All right.

403
00:28:04,760 --> 00:28:05,760
Thank you.