Building Smarter Organizations Through the Human-AI Partnership with McKinsey Senior Fellow Dr. Michael Chui

WEBVTT
1
00:00:00.091 --> 00:00:00.571
Hello, everyone.
2
00:00:00.591 --> 00:00:02.432
Welcome to another episode of Future Ready Leadership.
3
00:00:02.452 --> 00:00:04.232
My guest today, Dr. Michael Chewy.
4
00:00:04.672 --> 00:00:15.234
He is a senior fellow at Quantum Black AI by McKinsey, and he leads research on the impact of disruptive technologies and innovation on business, the economy and society.
5
00:00:15.835 --> 00:00:25.537
And he has led McKinsey research in a lot of different areas, all very relevant to what we're going to be talking about today, including AI, robotics, automation, future of work, data analytics,
6
00:00:26.157 --> 00:00:35.859
collaboration technologies, which I used to write about way, way back in the day, 14, 15 years ago, the Internet of Things, which we don't even hear about much anymore, and biological technologies.
7
00:00:36.219 --> 00:00:37.219
Michael, thank you for joining me.
8
00:00:37.780 --> 00:00:38.380
Great to be here.
9
00:00:38.420 --> 00:00:39.160
Thanks for having me on.
10
00:00:40.000 --> 00:00:48.462
So today, a few hours ago, before I jumped on this call with you, I had a meeting with my CHRO group called Future of Work Leaders.
11
00:00:49.764 --> 00:00:54.548
And one of the topics that came up was the subject of AI.
12
00:00:54.648 --> 00:00:57.190
As you can imagine, it comes up in a lot of conversations.
13
00:00:58.871 --> 00:01:03.975
And there were a few really interesting things that came up from these CHROs leading some very big global organizations.
14
00:01:05.636 --> 00:01:12.822
And I kind of wanted to just share some of the themes that came up and get your thoughts on it and see if, you know, we're moving in the right direction.
15
00:01:14.143 --> 00:01:18.867
So in a new book that I'm working on, I have this concept that I created called the algorithmic enterprise.
16
00:01:20.034 --> 00:01:23.578
And there are several, I guess you could say, potential futures that I've outlined.
17
00:01:24.459 --> 00:01:26.120
One of them is the algorithmic enterprise.
18
00:01:26.160 --> 00:01:36.351
And it's kind of a cautionary thing that I'm worried about, which is what happens when organizations will rely too much on AI and on the algorithm for everything.
19
00:01:37.171 --> 00:01:41.215
And I don't know if you remember, there was this Black Mirror episode that was out, I
20
00:01:44.138 --> 00:01:51.105
And the episode basically looked at what happens when in our personal lives, we act and behave in a certain way to optimize the algorithm.
21
00:01:51.826 --> 00:01:56.270
So you go to a store, you smile at everybody, you leave the person a really big tip.
22
00:01:56.570 --> 00:01:58.392
They kind of rate you and give you points.
23
00:01:59.153 --> 00:02:00.073
You get out of the cab.
24
00:02:00.093 --> 00:02:00.734
You're like, hello.
25
00:02:00.854 --> 00:02:08.581
Like everything is designed in a way to game the system so that you get points and the algorithm views you as being good.
26
00:02:10.012 --> 00:02:20.398
And part of me wonders what's going to happen and can that happen inside of our organizations where, for example, if you're evaluated based on how well you collaborate, I'm just going to join every meeting to try to game the system.
27
00:02:21.539 --> 00:02:24.941
I'm never going to take vacation because I don't want to have like a disengagement flag in there.
28
00:02:25.441 --> 00:02:31.145
I'm not going to speak up or say anything negative to my leader because I don't want to get a reputation as being a disgruntled employee.
29
00:02:32.296 --> 00:02:46.116
So the first part of the question is, are you worried or are you thinking about that in any capacity, this algorithmic enterprise where we are just working and trying to optimize the algorithm and forget about the humanity at work?
30
00:02:47.723 --> 00:02:50.946
So yes, I think it's a real valid concern.
31
00:02:50.986 --> 00:02:53.447
I mean, I have a lot of thoughts as you were describing it, right?
32
00:02:53.467 --> 00:03:01.594
And so, you know, there's a funny saying, which is, you know, the history doesn't repeat itself, but it rhymes.
33
00:03:01.974 --> 00:03:03.875
And so I think that's often true here, too.
34
00:03:03.955 --> 00:03:05.677
And so a number of different things.
35
00:03:05.997 --> 00:03:07.298
Yeah, is it a real concern?
36
00:03:07.358 --> 00:03:08.299
Absolutely, right?
37
00:03:08.319 --> 00:03:10.700
I mean, and particularly, you know, when you have...
38
00:03:11.601 --> 00:03:24.533
You know, systems that are monitoring you all the time, whether it's keystrokes or whether it's, you know, you mentioned Internet of Things, you know, where you are or, you know, listening to, you know, what you're, you know, our meetings now can be transcribed by AI and all those sorts of things.
39
00:03:24.593 --> 00:03:27.836
And so and then are you being evaluated all the time by these systems?
40
00:03:27.876 --> 00:03:28.756
And so I think that is...
41
00:03:29.837 --> 00:03:34.780
You have questions about your life work and how you feel about all of that.
42
00:03:35.941 --> 00:03:41.625
But then also part of me goes to the... A lot of things you just described are things we do already, right?
43
00:03:41.645 --> 00:03:50.991
Like if you're in a service profession where tipping is the custom, you do perform in a certain way hoping to get tips, for instance.
44
00:03:51.011 --> 00:03:57.074
When you're in front of your manager, not even speaking to them, you're kind of aware that they're looking at you and evaluating you.
45
00:03:57.114 --> 00:03:57.455
And so...
46
00:03:59.101 --> 00:04:00.722
To a certain extent, this already exists.
47
00:04:00.842 --> 00:04:13.788
And, you know, particularly with regard to AI, AI is so interesting, particularly the latest versions of AI, which, again, I have no, at least I don't believe currently that any AI has consciousness or real feelings.
48
00:04:14.628 --> 00:04:28.311
But I had the opportunity to talk with Ethan Mullick, a professor at Wharton, and he said to get these systems to, quote unquote, be more creative, to come up with better ideas, sometimes actually you treat it like a human being.
49
00:04:28.992 --> 00:04:34.513
And the same sorts of things you would say to a person will cause the AI to act sort of like a person would.
50
00:04:35.033 --> 00:04:43.295
And so, you know, in a certain sense, again, I don't think AI now is like a person, but the sorts of things you described happening are...
51
00:04:43.675 --> 00:04:47.157
are sometimes the things that you could use to quote-unquote game the system.
52
00:04:47.377 --> 00:04:50.759
But we also game the system with regard to human people too.
53
00:04:51.340 --> 00:05:01.606
And so do I think, yeah, are there ways in which this can be quite negative, quite negative for our experiences as people, etc.?
54
00:05:02.527 --> 00:05:03.827
Yeah, absolutely.
55
00:05:04.328 --> 00:05:06.669
Are these things we already do in some sense?
56
00:05:07.149 --> 00:05:07.890
Yes, too.
57
00:05:08.010 --> 00:05:11.192
And so that's where we're just in a realm of trying to figure out
58
00:05:11.872 --> 00:05:15.155
Look, what is, you know, ethical?
59
00:05:15.235 --> 00:05:16.416
What does provide integrity?
60
00:05:16.436 --> 00:05:19.499
What are the commitments we're making to one another in a work context?
61
00:05:19.979 --> 00:05:32.129
And that's, you know, technology is a lever that can just, you know, turn the knob of the power, the degree to which this is true, because now it can't observe, you know, listen to every single word that you say when you're, you know, talking to somebody on Zoom.
62
00:05:32.149 --> 00:05:32.269
Right.
63
00:05:33.470 --> 00:05:35.791
As an enterprise, what are we going to do with that?
64
00:05:36.712 --> 00:05:41.834
I suppose the difference is you – and I get what you're saying, kind of the parallel there.
65
00:05:41.854 --> 00:05:47.277
But today you kind of game the system in a way to create rapport with people.
66
00:05:47.437 --> 00:05:50.279
So you kind of game the system in a human way.
67
00:05:50.899 --> 00:05:56.222
But I'm worried about what happens when you game the system in a purely points –
68
00:05:57.162 --> 00:05:58.783
algorithmic driven way.
69
00:05:58.863 --> 00:06:04.087
So for example, and it's a totally lame, uh, lame, uh, version of this.
70
00:06:04.147 --> 00:06:08.170
So in, in my Tesla, there is a game called sky force.
71
00:06:09.593 --> 00:06:11.936
And anybody who has a Tesla knows of Sky Force.
72
00:06:12.016 --> 00:06:12.817
It's very addictive.
73
00:06:13.818 --> 00:06:20.405
And so one of the things that I recently realized is that when I start off the game, I don't know if there's like a glitch in the game or what's going on.
74
00:06:20.445 --> 00:06:28.634
But whenever I start the game, for the first 30, 45 seconds, it says you can play and you're going to get extra coins that you can use to upgrade your ship.
75
00:06:29.825 --> 00:06:42.709
And so what I realized is that if I play for a few minutes and then I exit out, I go to another game like Sudoku and then I exit out and I go back into Sky Force, it thinks that I am playing for the first time again and again it will give me extra coins.
76
00:06:43.550 --> 00:06:49.972
And so what happens is now my behavior, the way that I play that game, has changed to optimize...
77
00:06:51.037 --> 00:06:53.759
my getting more coins and more points.
78
00:06:54.679 --> 00:07:02.124
And so that's sort of what I'm worried about inside the organization, where if you are purely being evaluated based on an algorithm, right?
79
00:07:02.144 --> 00:07:08.387
Because we're talking about people analytics and data analytics and predictive analytics, and you're gonna get flagged for this, and what skills do you have?
80
00:07:08.427 --> 00:07:17.532
And so if the algorithm is gonna make those decisions and evaluate and judge us purely on those behaviors that we take, because the algorithm doesn't know...
81
00:07:18.513 --> 00:07:19.194
It can't tell.
82
00:07:19.694 --> 00:07:23.937
Like if I, if I see you at work and I say, Hey, Michael, how's it going?
83
00:07:24.798 --> 00:07:26.799
You can walk away from that and be like, yeah, that was total BS.
84
00:07:26.819 --> 00:07:27.479
That guy hates me.
85
00:07:27.519 --> 00:07:33.203
Like I get what he's doing, but an algorithm has no idea that I'm doing that.
86
00:07:34.344 --> 00:07:37.587
So an algorithm is going to say, wow, Jacob's joining all of these meetings.
87
00:07:37.627 --> 00:07:39.368
He must be a great collaborator.
88
00:07:39.388 --> 00:07:41.009
He must be building wonderful networks.
89
00:07:41.429 --> 00:07:44.551
Meanwhile, on the human level, everyone's like, what is this guy doing?
90
00:07:45.312 --> 00:07:46.353
This guy's so lame.
91
00:07:46.373 --> 00:07:46.933
I hate Jacob.
92
00:07:46.953 --> 00:07:47.554
Why is he here?
93
00:07:47.574 --> 00:07:47.674
Yeah.
94
00:07:48.480 --> 00:08:00.859
So that's kind of what I'm wondering if we'll get to that dystopian black mirror gaming the system at some point.
95
00:08:01.079 --> 00:08:02.261
I mean, it seems like it's possible.
96
00:08:02.702 --> 00:08:02.982
Yeah.
97
00:08:04.311 --> 00:08:05.873
I think these are choices, right?
98
00:08:05.893 --> 00:08:06.894
These are choices we can make.
99
00:08:06.934 --> 00:08:09.357
And again, to the point about like, is it happening already?
100
00:08:09.437 --> 00:08:09.638
Right.
101
00:08:09.698 --> 00:08:13.403
I mean, like as your CHRO folks, right?
102
00:08:14.344 --> 00:08:15.625
Sales comp, right?
103
00:08:16.086 --> 00:08:20.592
You have people, they're super attuned to the best way to meet the metrics.
104
00:08:20.772 --> 00:08:23.734
you know, as a Zillsbury, like super duper to do it.
105
00:08:23.894 --> 00:08:24.394
Oh gosh.
106
00:08:24.434 --> 00:08:25.075
People get creative.
107
00:08:25.575 --> 00:08:26.175
Yeah.
108
00:08:26.375 --> 00:08:32.639
And so I think one of the interesting things, I mean, I once had a conversation with someone who's, who is look in the field of education.
109
00:08:32.979 --> 00:08:33.180
Right.
110
00:08:33.740 --> 00:08:36.542
And, you know, I had this discussion about teaching to the test, right.
111
00:08:36.562 --> 00:08:37.522
It sounds wrong.
112
00:08:37.622 --> 00:08:37.802
Right.
113
00:08:38.263 --> 00:08:40.864
And, and they said something really provocative to me.
114
00:08:40.884 --> 00:08:41.785
They said, well,
115
00:08:42.105 --> 00:08:42.586
You know what?
116
00:08:43.367 --> 00:08:45.810
It's not so bad if the test is good.
117
00:08:46.230 --> 00:08:52.338
If the test actually reflects the things that you want the student to learn, then teaching the test is not so bad.
118
00:08:52.838 --> 00:08:58.726
And so to the algorithmic point, I think if you have algorithms that do reflect the
119
00:08:59.166 --> 00:09:09.891
you know, what we want from our employees, from our associates, whether it's higher sales or it's the way that they interact with other people or with customers or, you know, whatever the performance metrics are.
120
00:09:10.491 --> 00:09:16.714
If you have a good test, if the algorithm is measuring the things you really want, and it might be authenticity, which is harder,
121
00:09:17.988 --> 00:09:18.288
Good.
122
00:09:18.308 --> 00:09:18.989
All right.
123
00:09:19.469 --> 00:09:30.636
On the other hand, you can imagine either gamification or, you know, as you said, that, you know, encouraging behaviors and activities which are negative for people, customers, employees, etc.
124
00:09:30.656 --> 00:09:40.142
And so I think the onus is upon, you know, the people trying to evaluate, just like in our evaluations today, to measure the things that really matter.
125
00:09:41.353 --> 00:09:47.078
If you've ever asked yourself what good AI can really accomplish, well, you're not alone.
126
00:09:47.638 --> 00:09:53.203
And in fact, by some estimates, over 80% of AI projects fail because of poor data.
127
00:09:53.743 --> 00:09:55.024
But here's some good news for you.
128
00:09:55.505 --> 00:10:01.169
When AI leverages the rich, authentic data of employee recognition, you get something truly unique.
129
00:10:01.710 --> 00:10:07.595
Human intelligence from WorkHuman, creators of the world's number one recognition platform.
130
00:10:08.275 --> 00:10:13.921
Human intelligence drives good culture and engagement and unlocks insights into performance and skills.
131
00:10:14.521 --> 00:10:22.690
It also helps coach employees on what good peer-to-peer recognition looks like, turning every good job into better attention and well-being.
132
00:10:23.250 --> 00:10:27.655
And it helps surface feel-good stories that illustrate the best of your company culture.
133
00:10:28.255 --> 00:10:33.940
So in reality, AI can accomplish a whole lot of good with the power of human intelligence.
134
00:10:34.421 --> 00:10:42.287
To learn how today's leading companies are turning AI into a force for good with human intelligence, go to workhuman.com.
135
00:10:42.748 --> 00:10:45.750
Again, that is workhuman.com.
136
00:10:48.313 --> 00:10:55.419
It seems like we do need to then, I don't know if you would call it training or just awareness training.
137
00:10:56.822 --> 00:11:02.145
but I guess we need to start having conversations inside of our organizations on over-reliance of AI.
138
00:11:03.526 --> 00:11:12.612
So kind of another example that came up, one of the organizations in our group, which I won't name, they're saying, oh, we're giving chat GPT to everybody, right?
139
00:11:12.652 --> 00:11:14.833
Tens of thousands of employees, everyone's getting chat GPT.
140
00:11:16.354 --> 00:11:18.815
And I said, well, that's interesting.
141
00:11:18.835 --> 00:11:23.979
But what happens if you're preparing for a meeting, and let's say it's a new strategic direction,
142
00:11:25.104 --> 00:11:30.027
You go into ChatGPT, you feed all this data in there and you say, hey, I'm very for this new strategic direction.
143
00:11:31.768 --> 00:11:38.132
Make me a bulletproof argument for why we should go in this direction so that I can present it to my team.
144
00:11:39.292 --> 00:11:41.734
And then you have somebody else who also has access to ChatGPT.
145
00:11:42.074 --> 00:11:43.095
They don't like this strategy.
146
00:11:43.775 --> 00:11:50.099
And they go in there and they say, make me a bulletproof argument for why this strategic direction is bad.
147
00:11:51.113 --> 00:11:55.837
And now you have two people in there who are basically arguing chat GPT points.
148
00:11:57.198 --> 00:12:09.787
And so the other kind of scenario that I'm wondering about is what happens when humans basically become the regurgitators of information that AI tools give them.
149
00:12:10.768 --> 00:12:13.130
Because you won't know, right?
150
00:12:13.150 --> 00:12:17.153
I mean, in a meeting, you're not going to know where these people are getting information from.
151
00:12:18.133 --> 00:12:31.021
But if everyone is now using ChatGPT for emails, for performance reviews, for preparing for meetings, you're kind of just like you are now the puppet of ChatGPT and it's telling you what to say.
152
00:12:31.321 --> 00:12:31.782
You know what I mean?
153
00:12:31.802 --> 00:12:32.882
Like you are now the bot.
154
00:12:33.563 --> 00:12:37.105
And the ChatGPT is now the human that's telling you what to do.
155
00:12:37.125 --> 00:12:39.406
Like it's weird, right?
156
00:12:40.087 --> 00:12:43.529
But that's also something that I'm wondering if that could happen.
157
00:12:44.929 --> 00:12:45.970
I think it can happen.
158
00:12:45.990 --> 00:12:47.212
And a couple of things, right?
159
00:12:47.272 --> 00:12:49.954
There's a little bit about how do you view these technologies.
160
00:12:50.895 --> 00:12:58.163
And, you know, there's recently been a paper by some folks who talk about AI being a normal technology.
161
00:12:58.771 --> 00:13:06.716
And so I thought, you know, one thing that I thought you were going is, you know, for instance, you know, what if you're preparing for a meeting and suddenly you lose access to that technology?
162
00:13:06.816 --> 00:13:08.237
Are you able to do anything right?
163
00:13:08.717 --> 00:13:15.721
Do your skills for critical thinking, for creativity, et cetera, decline because you become used to these technologies?
164
00:13:15.741 --> 00:13:22.785
And I think there's some evidence that that might be true, but it's, you know, it's about wayfinding, you know, maps or not having maps and those sorts of things.
165
00:13:23.085 --> 00:13:25.767
But, yeah, there's some possibility that might be true as well.
166
00:13:25.787 --> 00:13:26.728
But on the other hand...
167
00:13:27.688 --> 00:13:29.528
I feel that way about Excel, right?
168
00:13:29.769 --> 00:13:36.590
If I lose access to Excel before a meeting and I need to do some quantitative analysis or whatever tool that I'm using at the time, that's a problem, right?
169
00:13:36.630 --> 00:13:40.951
And so to a certain extent, there is an echo of the past there.
170
00:13:41.211 --> 00:13:47.492
But also to your point about, do we end up being just the way that bots communicate with one another?
171
00:13:47.532 --> 00:13:50.353
I mean, that's probably going to happen without us to a certain extent.
172
00:13:51.293 --> 00:13:55.894
You know, there'd be bots, you know, that answer customer service requests.
173
00:13:56.014 --> 00:13:59.535
And I might have my own, you know, personal assistant bot.
174
00:13:59.555 --> 00:14:03.596
And the bots talk to one another and negotiate on my behalf and the company's behalf.
175
00:14:03.916 --> 00:14:04.056
Right.
176
00:14:04.156 --> 00:14:07.757
All of these things, you might think of them as dystopia.
177
00:14:08.537 --> 00:14:09.938
I think they're coming at some sense.
178
00:14:09.958 --> 00:14:10.078
Right.
179
00:14:10.098 --> 00:14:11.098
And so it's going to get really complicated.
180
00:14:11.118 --> 00:14:15.059
I think you're saying, oh, this with a smile and such a cheerful demeanor and a happy face.
181
00:14:16.825 --> 00:14:17.886
I live in sunny California.
182
00:14:17.906 --> 00:14:18.526
I'm an optimist.
183
00:14:18.546 --> 00:14:20.507
But yeah, there are things that are hard about that.
184
00:14:20.608 --> 00:14:22.529
And they can be really problematic about that.
185
00:14:22.989 --> 00:14:34.357
And so, again, I think the one interesting analogy, if you're talking about a strategy, for instance, is that just think about our most highest level executives.
186
00:14:34.457 --> 00:14:37.239
When they go into a meeting, they're prepped.
187
00:14:37.499 --> 00:14:41.562
But they're often prepped, not necessarily with ChatGPT, but by staffs.
188
00:14:42.528 --> 00:14:49.612
And you could take that sort of view, oh, are these just two staffs talking to one another through an executive?
189
00:14:50.252 --> 00:14:52.274
And I think the answer there is no.
190
00:14:52.394 --> 00:14:52.654
Why?
191
00:14:52.714 --> 00:14:57.877
Because the executive is involved, hopefully, if they're a good executive, in the process.
192
00:14:58.557 --> 00:15:05.541
And so presumably, if I'm using ChatGPT as part of my staff, I'm not not thinking.
193
00:15:05.841 --> 00:15:07.302
I hope I'm not thinking when I'm doing it.
194
00:15:07.322 --> 00:15:07.642
I hope so too.
195
00:15:07.662 --> 00:15:08.603
I'm having a conversation.
196
00:15:08.663 --> 00:15:10.524
I might be having a conversation with people and...
197
00:15:11.000 --> 00:15:13.041
you know, AI, et cetera.
198
00:15:13.541 --> 00:15:21.285
And then when I come to the meeting, because gosh, if it's worth having a meeting, I am there, you know, then I'm informed by all of these things.
199
00:15:21.825 --> 00:15:23.486
And so I don't think that's dystopian.
200
00:15:23.706 --> 00:15:26.848
I mean, I think that is, you know, adding to my capability.
201
00:15:26.868 --> 00:15:35.372
Sometimes we talk about AI, you know, creating superpowers and it's like part of my staff, you could say could be, you know, these AI systems.
202
00:15:36.032 --> 00:15:37.113
And I think that's a good thing.
203
00:15:37.473 --> 00:15:40.034
I'm thinking of it kind of like, it's like a silent automation, right?
204
00:15:40.074 --> 00:15:44.476
So many people are worried about, um, AI and automation taking away jobs.
205
00:15:45.177 --> 00:15:53.741
I'm more worried about it, not taking away jobs, but silently automating people in terms of their cognitive capacity where they become the mouthpieces for AI.
206
00:15:54.721 --> 00:16:03.966
And it's funny because I, um, like I, I shared the story in our, in our WhatsApp, in our, uh, CHRO group, but I'm a, on a WhatsApp group with, with some friends.
207
00:16:05.268 --> 00:16:08.350
And, you know, people were debating and going back and forth on something.
208
00:16:08.630 --> 00:16:13.594
And, you know, when you go into ChatGPT and you give it a prompt, it gives you the response.
209
00:16:13.654 --> 00:16:18.877
But at the very end, it tells you, here is what you wanted with a soft and whatever tone that meets blah, blah, blah.
210
00:16:18.897 --> 00:16:20.378
Like it explains to you what it just did.
211
00:16:21.499 --> 00:16:23.060
And so we're in this WhatsApp conversation.
212
00:16:23.100 --> 00:16:23.400
I don't know.
213
00:16:23.420 --> 00:16:25.081
There's like 20 people in there.
214
00:16:26.062 --> 00:16:28.003
And this person puts in a response.
215
00:16:28.564 --> 00:16:31.986
And he forgets to delete at the very end the ChatGPT piece.
216
00:16:33.633 --> 00:16:49.562
And now everybody's trolling this person like crazy, but I was thinking about that and I'm like, what is happening in the world where you can no longer have a conversation with your friends in a WhatsApp group without going into ChatGPT?
217
00:16:49.962 --> 00:16:51.163
And you could see the thing that he said.
218
00:16:51.203 --> 00:16:57.547
He wanted something that was warm and neutral and friendly that didn't cause any arguments or debates.
219
00:16:57.607 --> 00:17:00.989
And it's kind of like, it really just made me wonder,
220
00:17:03.060 --> 00:17:13.606
Are you no longer a person if you have to go into WhatsApp or into ChatGPT and curate a response to friends?
221
00:17:14.647 --> 00:17:27.815
But then similarly, there are also lots of stories, whether it's a kid's fundraiser or where you're writing a marketing campaign or a landing page where you want to go into something like ChatGPT and ask it to write you something very persuasive to get people to click.
222
00:17:29.155 --> 00:17:32.017
So the use cases really matter of like,
223
00:17:33.340 --> 00:17:34.960
where it's important to be human.
224
00:17:34.980 --> 00:17:50.865
And, and I don't know if that's something that you can teach or if it's just morality or ethics, but it's, you know, like, that's really what hit me when I had a friend of mine that was using chat GPT in a WhatsApp conversation.
225
00:17:51.285 --> 00:17:52.425
Yeah.
226
00:17:52.845 --> 00:17:58.407
I think this is one of those things that, you know, we individually, and as a society and as friend groups,
227
00:17:59.632 --> 00:18:06.614
we'll adapt and learn to this, you know, learn how to use this technology and, and our norms will develop over time.
228
00:18:06.894 --> 00:18:16.116
I'll tell you, you know, I don't know, there, there are some times when it's so hard for me to express what I want to express.
229
00:18:16.897 --> 00:18:18.497
I don't know, condolences, for instance.
230
00:18:20.698 --> 00:18:25.059
And it's, it's searchable on the web, right?
231
00:18:25.099 --> 00:18:27.099
Like how, how do I tell this person?
232
00:18:27.520 --> 00:18:29.040
And it's been helpful to me.
233
00:18:29.894 --> 00:18:34.918
And, you know, can I imagine, you know, asking ChatGPT to help with that too?
234
00:18:35.258 --> 00:18:35.739
Yes.
235
00:18:36.619 --> 00:18:40.622
But at the end of the day, what I write is I'm responsible for that.
236
00:18:40.923 --> 00:18:46.307
Also, I will ask a friend or, you know, ask my spouse, hey, does this say what I mean it to say?
237
00:18:47.087 --> 00:18:50.070
And if I think about it that way, yeah.
238
00:18:50.750 --> 00:18:55.474
But to your point, right, like people joking around in a WhatsApp group or something like that and someone feels, eh.
239
00:18:56.855 --> 00:18:58.896
But as you said, you kind of like roasted them, right?
240
00:18:58.916 --> 00:19:00.236
So like, you know.
241
00:19:00.416 --> 00:19:03.057
Well, because it's important as a human being to make mistakes.
242
00:19:03.318 --> 00:19:04.918
It's important to say the wrong thing.
243
00:19:05.038 --> 00:19:06.779
It's important sometimes to look stupid.
244
00:19:06.839 --> 00:19:10.200
It's like, that's just the nature of it, right?
245
00:19:10.220 --> 00:19:10.620
Yeah.
246
00:19:10.940 --> 00:19:19.444
But if you try to create every email, every conversation, every text in a perfect way, you lose the humanness.
247
00:19:20.424 --> 00:19:22.765
And so part of what I'm
248
00:19:24.002 --> 00:19:34.330
I don't know if fear is the right word, but part of me that I'm trying to understand is if we get too reliant on AI and these technologies that we lose the humanness.
249
00:19:34.590 --> 00:19:43.696
It's sort of like, I think there was a study that was done where if you rely so much on GPS when it comes to navigation that there's literally a part of your brain that actually dies.
250
00:19:44.517 --> 00:19:52.883
And so there are a lot of people now, and I've talked to friends like this, where even if they're driving two miles from home and they know exactly where they're going, they have the GPS on.
251
00:19:53.902 --> 00:19:56.846
Like they can't navigate the world anymore.
252
00:19:56.966 --> 00:19:59.349
And that to me is not good.
253
00:20:00.109 --> 00:20:00.250
Right?
254
00:20:00.270 --> 00:20:03.894
I mean, even now when I go to a new city or a new country, I quickly look at a map.
255
00:20:05.075 --> 00:20:09.020
I never turn on GPS and I just wander and try to get my bearings and figure out where I'm going.
256
00:20:10.099 --> 00:20:27.489
And so if people are so reliant just to get a few miles from home using GPS, what's going to happen in the context where we're scared to have conversations, we're scared to go to meetings, we're scared to do a performance review, we're scared to do anything unless we have this like AI assistant that, you know, tell me what to do.
257
00:20:28.070 --> 00:20:28.850
I don't know what to say.
258
00:20:29.531 --> 00:20:32.312
Like that to me is weird, right?
259
00:20:32.613 --> 00:20:36.435
And I don't want to feel like I'm having conversations with a bot that
260
00:20:37.383 --> 00:20:42.748
Like I know my friend is saying it, but I know it's a bot who created it.
261
00:20:43.249 --> 00:20:44.830
It's like the uncanny valley, right?
262
00:20:46.052 --> 00:20:46.832
I don't know.
263
00:20:47.233 --> 00:20:48.214
Something about that freaks me out.
264
00:20:48.922 --> 00:20:51.785
I think we're going to figure it out over time is what I think, right?
265
00:20:51.805 --> 00:20:58.151
Like the wayfinding thing is the reference I was making before about, you know, having the GPS on.
266
00:20:58.191 --> 00:21:01.154
Although there's some value if it shows how much traffic there is.
267
00:21:01.634 --> 00:21:02.115
Yes.
268
00:21:02.155 --> 00:21:02.495
Yeah.
269
00:21:02.735 --> 00:21:06.058
I think we'll find our way through this as people start, right?
270
00:21:06.078 --> 00:21:10.202
Because even on social media, you know, people using these apps to change the way they look, right?
271
00:21:10.283 --> 00:21:10.543
I mean...
272
00:21:12.173 --> 00:21:16.716
But we're figuring out, at least the kids who use it most are figuring that out, right?
273
00:21:17.097 --> 00:21:21.040
And you're not as surprised when you show up and somebody looks slightly different and all those sorts of things.
274
00:21:21.220 --> 00:21:25.263
You can tell now when somebody uses a filter on Instagram, you're like, nah, you don't look like that.
275
00:21:26.604 --> 00:21:27.084
I know you.
276
00:21:27.164 --> 00:21:27.765
I've seen you.
277
00:21:27.825 --> 00:21:28.786
That's not what you look like.
278
00:21:29.506 --> 00:21:30.747
But we're going to start doing this.
279
00:21:30.767 --> 00:21:34.070
Like, you know, I rarely have spelling mistakes anymore.
280
00:21:34.370 --> 00:21:39.054
You know, does that make me less human because I'm less vulnerable because the system told me that's not the right spelling, right?
281
00:21:39.074 --> 00:21:39.214
Yeah.
282
00:21:40.110 --> 00:21:40.870
You know, I don't know.
283
00:21:40.950 --> 00:21:48.614
And I think our norms and our expectations and our literacy with the technology, you know, change over time.
284
00:21:49.134 --> 00:22:00.319
Is it particularly weird because these technologies can do things like provide advice about emotional or, you know, expressive things that we think are very inherently human?
285
00:22:00.359 --> 00:22:08.242
Yeah, I think it's a little weird, but we're, I think we're going to go figure it out as we go along because now we, you know, I had this experience, um,
286
00:22:09.874 --> 00:22:10.355
What was it?
287
00:22:10.475 --> 00:22:16.303
Oh, I had emailed a colleague, I think with a job reference or a suggestion to somebody.
288
00:22:16.684 --> 00:22:20.629
And I got a text from him and he says, Michael, did you send me that email?
289
00:22:21.210 --> 00:22:22.152
I said, yeah.
290
00:22:22.292 --> 00:22:23.894
And he texted me, was it legit?
291
00:22:24.154 --> 00:22:25.577
I was like, yeah.
292
00:22:26.351 --> 00:22:29.332
What, did you think I, you know, what do you think it was ChatGPT?
293
00:22:29.392 --> 00:22:30.112
And he goes, yeah.
294
00:22:30.373 --> 00:22:31.193
I was like, well, thanks.
295
00:22:31.553 --> 00:22:32.193
Thanks, Rob.
296
00:22:32.653 --> 00:22:37.775
You know, but again, that's just part of this process of our figuring out, hey, you know, this is the norm.
297
00:22:37.795 --> 00:22:38.736
This is not the norm.
298
00:22:38.776 --> 00:22:39.576
That felt weird.
299
00:22:39.736 --> 00:22:40.536
All those sorts of things.
300
00:22:40.556 --> 00:22:41.197
We're going to do that.
301
00:22:42.537 --> 00:22:47.339
What do you, like if you were to imagine, I don't know, three, four or five years, right?
302
00:22:47.379 --> 00:22:55.382
Because we're even talking about, we didn't even talk about agentic AI, which we'll talk about in a minute, but there's also kind of artificial general intelligence that we're talking about.
303
00:22:56.038 --> 00:23:02.000
I mean, I'm imagining we're what on ChatGPT 4.5, ChatGPT 5, ChatGPT 6, Grok 7.
304
00:23:03.560 --> 00:23:06.101
How much better these tools are going to be?
305
00:23:06.121 --> 00:23:13.764
Do you, like, how are we going to balance the humanity and the technology piece?
306
00:23:14.304 --> 00:23:17.005
I know you say, oh, you know, we'll figure it out, but are there.
307
00:23:18.350 --> 00:23:19.331
What do you think it's going to take?
308
00:23:19.711 --> 00:23:20.451
Is it training?
309
00:23:20.531 --> 00:23:21.752
Is it conversations?
310
00:23:21.892 --> 00:23:27.695
Is it like what do we need to do to find that balance?
311
00:23:27.755 --> 00:23:30.456
Practice, I think, is the one word that I would use.
312
00:23:30.897 --> 00:23:31.517
Okay.
313
00:23:31.817 --> 00:23:38.080
And, you know, we will develop practices and that sort of, and we will be more practiced at using these technologies.
314
00:23:38.100 --> 00:23:41.022
By the way, I mean, depending on your point of view, I mean, there's some people think,
315
00:23:41.581 --> 00:23:43.963
You might have seen this recent, you know, AI 2027.
316
00:23:44.003 --> 00:23:48.888
I mean, some people think things will move very, very quickly over the next couple of years versus five years.
317
00:23:49.968 --> 00:23:51.570
But I think adoption takes longer than that.
318
00:23:51.930 --> 00:23:54.773
But nevertheless, I mean, I think those, but, you know...
319
00:23:55.482 --> 00:24:05.767
To your point about ethics or morality, you know, we do give in many cases or many of us primacy to people.
320
00:24:06.207 --> 00:24:12.770
And so, you know, it changes the way I interact when I know I'm talking to Jacob.
321
00:24:13.410 --> 00:24:16.133
versus I am talking to a bot, at least for now.
322
00:24:17.614 --> 00:24:18.535
And that matters.
323
00:24:18.675 --> 00:24:32.647
Even if I'm using language, even if I'm speaking and listening, all those things being true, the sort of surface-y, just at least currently, it matters to me when I'm talking to a person.
324
00:24:32.967 --> 00:24:38.171
And so I think that underpins a lot of the things you're talking about.
325
00:24:38.352 --> 00:24:39.192
What does it mean to be human?
326
00:24:40.395 --> 00:24:45.538
For me to be human means when I talk to a human, I think about that differently.
327
00:24:45.578 --> 00:24:46.818
I feel about that differently.
328
00:24:48.039 --> 00:24:59.205
I just wonder if in the coming years, there's going to be kind of like this battle between, it's almost like the battle between, you know, people who want EVs versus people who are like, I want the traditional muscle car, right?
329
00:24:59.285 --> 00:25:04.428
So people who are going to be like, I'm purely human versus like, no, I'm using human with AI.
330
00:25:04.448 --> 00:25:08.550
Like if there's going to be this dual of how pure of a human do you want to be?
331
00:25:09.447 --> 00:25:10.807
It's even like the movie, The Matrix, right?
332
00:25:10.827 --> 00:25:17.769
There are some people who were born in the matrix and there are other people who are just kind of born as humans who don't have the plugs on them.
333
00:25:19.189 --> 00:25:20.529
So time will tell on that.
334
00:25:21.430 --> 00:25:26.931
But I wanted to transition a little bit to the concept of agentic AI, because this is obviously a huge topic of conversation.
335
00:25:26.951 --> 00:25:32.292
And we had a CHRO meetup earlier in the year.
336
00:25:32.312 --> 00:25:33.712
I think it was in January.
337
00:25:34.072 --> 00:25:38.173
And one of our member companies mentioned something that was really interesting.
338
00:25:39.123 --> 00:25:49.652
And she said that when they are looking to hire new talent, like they bring in their CTO, who's a part of the meetings with the CHRO and, you know, the people.
339
00:25:50.953 --> 00:25:59.920
And during the interview process, it's up to the CEO to really weigh in on the hiring decisions because part of his role is to say, can this job be done with AI?
340
00:26:00.161 --> 00:26:01.562
Do we need this person?
341
00:26:01.582 --> 00:26:01.722
Right.
342
00:26:03.075 --> 00:26:14.146
And I thought that was very interesting to have somebody on the technical side involved in the interview process to really decide, do we need this person to be here or can this be done with an agent?
343
00:26:14.206 --> 00:26:15.547
Can this be done with technology?
344
00:26:16.228 --> 00:26:18.190
And saying this is a very interesting change.
345
00:26:19.291 --> 00:26:25.797
We, we usually don't think about technology being involved in the hiring process of like sitting in on the interviews.
346
00:26:26.758 --> 00:26:26.978
So.
347
00:26:27.880 --> 00:26:31.902
What do you think this is going to mean for the future of talent?
348
00:26:31.962 --> 00:26:41.087
And maybe even before we get into that, can you take one step back, explain what agentic AI is, and then maybe we can jump into this talent piece there.
349
00:26:41.518 --> 00:26:41.738
Yeah.
350
00:26:42.499 --> 00:26:42.859
It's funny.
351
00:26:42.879 --> 00:26:54.165
When I talk about agentic AI, oftentimes I've joked, it sounds like when a parent is bragging about their kid and the first thing they have to explain is why it's hard, why the thing that they're so proud of is hard.
352
00:26:54.546 --> 00:26:59.989
Because a lot of times when I explain agentic AI, people say, I thought that was just AI, but I will explain it nevertheless, right?
353
00:27:00.009 --> 00:27:01.030
Because what?
354
00:27:01.470 --> 00:27:04.612
So number one, agentic, you know, acting in the world, right?
355
00:27:04.672 --> 00:27:06.353
So it's not just...
356
00:27:07.094 --> 00:27:16.638
Not just, you know, a chat bot, you can chat, but it doesn't, you know, it can't actually, you know, book a plane ticket for you or, you know, or do a financial transaction.
357
00:27:16.658 --> 00:27:17.238
So that's one.
358
00:27:17.799 --> 00:27:22.681
Two is it's based on modern AI, which are, you know, foundation models or large neural networks.
359
00:27:24.280 --> 00:27:31.062
And thirdly, it does things oftentimes multi-step relatively autonomously.
360
00:27:31.122 --> 00:27:38.665
So it's extended sets of actions powered by foundation models or modern AI acting in the world.
361
00:27:39.645 --> 00:27:43.566
And again, a lot of people will say, because all of those things have happened before, right?
362
00:27:43.586 --> 00:27:46.807
I mean, we have systems that have acted in the world forever, right?
363
00:27:46.867 --> 00:27:49.448
You know, your car has computers in it.
364
00:27:49.568 --> 00:27:50.188
And so that happens.
365
00:27:50.679 --> 00:27:55.742
We've had foundation models in AI, those underpin, you know, chat GPT, Claude and all those sorts of things.
366
00:27:55.762 --> 00:27:56.763
So that happens as well.
367
00:27:57.163 --> 00:27:59.364
And then we've had systems that are relatively autonomous.
368
00:28:00.485 --> 00:28:05.528
But when you had those systems, they often had to be programmed, you know, step by step by step by step.
369
00:28:06.329 --> 00:28:10.671
And that was expensive and usually had this, you know, long tail of exceptions.
370
00:28:10.711 --> 00:28:11.592
You know, when you're,
371
00:28:12.345 --> 00:28:15.527
when you're on and you're please press pound if you need to speak to a human, right?
372
00:28:15.587 --> 00:28:16.727
And all those sorts of things.
373
00:28:16.788 --> 00:28:16.988
Right.
374
00:28:17.328 --> 00:28:26.133
And the nice thing about foundation models or the remarkable thing about them is they can often deal with a whole set of exceptions that they weren't trained explicitly to deal with.
375
00:28:26.193 --> 00:28:26.353
Right.
376
00:28:26.373 --> 00:28:28.614
Because you train them with huge amounts of data.
377
00:28:28.634 --> 00:28:30.535
And so suddenly, Hey, guess what?
378
00:28:30.575 --> 00:28:32.116
I can ask it to write my,
379
00:28:33.216 --> 00:28:35.978
resume in iambic pentameter or whatever.
380
00:28:36.718 --> 00:28:38.459
No one trained it to do that, but it learned it.
381
00:28:38.960 --> 00:28:40.781
And so as a result, what happens?
382
00:28:40.821 --> 00:28:56.990
If you have an AI agent that can do things relatively autonomously and deal with, then you can start thinking about a customer service agent, which not only can follow the script, but can respond to a broader range of things.
383
00:28:57.541 --> 00:29:03.592
intelligibly than the press one, press two, or just have a menu tree.
384
00:29:04.234 --> 00:29:06.538
So that's agentic AI.
385
00:29:08.361 --> 00:29:08.502
Hmm.
386
00:29:09.723 --> 00:29:14.725
So it sounds like the difference between AI and agentic as the action piece.
387
00:29:15.305 --> 00:29:33.032
So being able to go from, like you said, a chat bot to book a flight, take some sort of an action, solve a problem that would require access to either data or systems or, I don't know, infrastructure, something that allows it to do something.
388
00:29:34.204 --> 00:29:35.504
We sometimes call it tool use.
389
00:29:35.825 --> 00:29:36.825
Tool use, yeah, yeah.
390
00:29:36.845 --> 00:29:37.625
That's a good way to think about it.
391
00:29:37.665 --> 00:29:37.885
Okay.
392
00:29:39.086 --> 00:29:50.730
So then what do you think about the story that I shared with you about the CTO or having a technical role involved in that hiring process to decide if this person is needed?
393
00:29:50.770 --> 00:29:55.852
Because it seems like instead of an interview process, it's now sort of like a candidate process.
394
00:29:55.932 --> 00:30:00.913
fighting for their life to like, yes, I, this is why you need me.
395
00:30:00.953 --> 00:30:03.134
This is why you need a person and a bot can't do it.
396
00:30:03.634 --> 00:30:05.615
It reminds me of the scene from, um, what was it?
397
00:30:05.695 --> 00:30:11.056
Office space where those, those two guys go in there and they're like, what exactly do you do here?
398
00:30:11.496 --> 00:30:14.257
And, and, uh, the guy's not able to explain it.
399
00:30:14.277 --> 00:30:16.798
And finally he gets frustrated and he's like, I'm a people person.
400
00:30:16.878 --> 00:30:17.238
Damn it.
401
00:30:17.418 --> 00:30:18.358
I deal with people.
402
00:30:19.318 --> 00:30:19.498
Right.
403
00:30:19.518 --> 00:30:21.679
So it reminds me of, of, of that.
404
00:30:23.199 --> 00:30:24.600
Office space will never die.
405
00:30:24.640 --> 00:30:25.620
It is absolutely true.
406
00:30:26.644 --> 00:30:28.146
That's our future that we're all destined for.
407
00:30:28.186 --> 00:30:28.826
We just don't know it.
408
00:30:29.467 --> 00:30:32.370
We'll all be like beating up the printer out in the field now.
409
00:30:32.710 --> 00:30:43.261
But I more often hear this dialogue and, you know, I think a lot of folks will have heard about the memo that came from a CEO.
410
00:30:44.342 --> 00:30:49.808
I think that dialogue from what I've heard happens more, less at the,
411
00:30:50.373 --> 00:30:53.737
interviewing stage, but more at the request for headcount stage.
412
00:30:54.077 --> 00:30:55.158
Okay.
413
00:30:55.959 --> 00:31:02.627
Because it's, I mean, it is really interesting if you're, you know, evaluating an individual against whether or not a machine can do what they can do.
414
00:31:03.127 --> 00:31:10.355
But more broadly, if, if a manager is saying, you know, annual plan or, you know, and trying to add additional headcount,
415
00:31:10.941 --> 00:31:20.691
And the challenge coming from CFO, CTO, CEO is, yes, but, you know, why can't this be done, you know, through some sort of machine?
416
00:31:21.291 --> 00:31:28.178
Now, with that said, again, the echo from history is we do already ask that question in our organizations, right?
417
00:31:28.238 --> 00:31:30.420
If I'm, you know, head of manufacturing,
418
00:31:30.907 --> 00:31:38.655
and I say I need to add headcount, it would be crazy for my superior not to ask, yes, but what about machines?
419
00:31:40.157 --> 00:31:42.740
And so I think, again, that echo is going to be there.
420
00:31:42.760 --> 00:31:48.666
But the interesting thing about AI, and this is borne out through our research as well, right?
421
00:31:48.686 --> 00:31:50.348
I mean, heretofore...
422
00:31:51.385 --> 00:31:55.810
The labor economists call it skill-biased technological change.
423
00:31:55.850 --> 00:31:58.153
I hate this term, high skill versus low skill.
424
00:31:58.193 --> 00:32:09.365
But largely, the technologies could automate the tasks of people who are lowest paid or for which their jobs have the lowest educational requirements.
425
00:32:09.585 --> 00:32:09.786
Yeah.
426
00:32:10.266 --> 00:32:12.591
Generative AI is the exact opposite.
427
00:32:12.871 --> 00:32:17.119
That has the most impact on jobs that require graduate degrees.
428
00:32:17.821 --> 00:32:21.467
It has the most impact on the jobs that we pay people to do the most.
429
00:32:21.507 --> 00:32:22.289
And so suddenly
430
00:32:23.328 --> 00:32:28.672
you know, maybe the political economy of it is the people in power realize, oh, no, now this stuff affects me, too.
431
00:32:29.252 --> 00:32:30.653
So, again, I think there's an echo.
432
00:32:30.973 --> 00:32:37.278
And it's, you know, we've had these types of discussions before, but now they're being applied to roles like sales.
433
00:32:37.298 --> 00:32:39.099
They're being applied to roles like engineering.
434
00:32:39.440 --> 00:32:42.662
They're being applied to these cognitive, quote, unquote, white collar roles.
435
00:32:43.322 --> 00:32:48.006
And so, again, is that challenging, disruptive?
436
00:32:48.026 --> 00:32:48.406
Yeah.
437
00:32:48.746 --> 00:32:49.868
scary in some cases.
438
00:32:49.968 --> 00:32:51.390
Yeah, it absolutely is.
439
00:32:52.171 --> 00:33:01.404
But it is an echo of things that within our organizations, within the people that we employ, it's not as if we haven't had these types of discussions before.
440
00:33:03.454 --> 00:33:04.535
Here's a good question for you.
441
00:33:05.076 --> 00:33:07.799
Who in your marketing department is a flight risk?
442
00:33:08.400 --> 00:33:11.243
How about where are your talent or skills gaps?
443
00:33:11.784 --> 00:33:14.587
Or which employees make up your next generation of leaders?
444
00:33:14.968 --> 00:33:16.770
Or maybe how do we build engagement?
445
00:33:17.391 --> 00:33:22.997
If you couldn't answer any of these, you need to check out Human Intelligence from WorkHuman.
446
00:33:23.538 --> 00:33:37.467
By combining AI with the rich data of their number one rated employee recognition platform, WorkHuman's human intelligence unlocks insights and capabilities that redefine talent management, cultural transformation, and employee engagement.
447
00:33:37.987 --> 00:33:42.150
In short, it'll give you a good answer to some of HR's toughest questions.
448
00:33:42.750 --> 00:33:50.295
To learn how today's leading companies are turning AI into a force for good with human intelligence, go to WorkHuman.com.
449
00:33:50.675 --> 00:33:53.037
Again, that is WorkHuman.com.
450
00:33:55.158 --> 00:33:57.300
What are you advising organizations?
451
00:33:57.340 --> 00:34:02.064
Obviously, you and McKinsey more broadly, you work with a lot of organizations.
452
00:34:02.084 --> 00:34:03.265
I know you do a lot of reports.
453
00:34:04.226 --> 00:34:08.589
You probably have a lot of executives who are coming to you saying, hey, Michael, tell me, what do I do?
454
00:34:08.629 --> 00:34:09.610
What should I be thinking about?
455
00:34:10.451 --> 00:34:17.877
So for people who are listening to this, and I'm not even sure how to frame the right question because AI is impactful in so many different areas, right?
456
00:34:19.613 --> 00:34:30.711
So how do you, I don't know, figure out the best course of action or if you were to give people advice on how to have the right conversations with AI, where would you point them?
457
00:34:30.771 --> 00:34:34.498
How do you start formulating your ideas and strategy around this?
458
00:34:34.913 --> 00:34:35.153
Yeah.
459
00:34:35.814 --> 00:34:38.576
I mean, I think there are a lot of different questions you can ask, right?
460
00:34:38.596 --> 00:34:43.881
I mean, you could have this, you know, employee experience question, sort of where we started our conversation.
461
00:34:43.901 --> 00:34:48.024
But let me take things in a different sort of strategic set of questions.
462
00:34:48.044 --> 00:34:51.447
And again, there are all kinds of ways in which AI can make a difference.
463
00:34:51.887 --> 00:34:55.771
But, you know, maybe some of our CHRO discussions have been about...
464
00:34:57.354 --> 00:35:00.497
the key question being, what is our organization of the future?
465
00:35:00.897 --> 00:35:04.461
What will our enterprise look like five years from now?
466
00:35:04.982 --> 00:35:06.763
We know the shape of it now.
467
00:35:06.824 --> 00:35:09.466
We know how many our employees are in different roles.
468
00:35:09.927 --> 00:35:14.131
We have some sense of the skills they need and what it takes to be successful.
469
00:35:14.151 --> 00:35:16.794
A question that we are often asked, the organization, the organization, the organization,
470
00:35:17.138 --> 00:35:19.782
yeah, but what will our organization look like five years from now?
471
00:35:20.262 --> 00:35:23.066
And then how do we get there, right, from the from-to?
472
00:35:23.647 --> 00:35:28.854
And one of the things that informs the from-to, as, you know, to your CTO point, is the technology.
473
00:35:29.434 --> 00:35:32.038
What will technology be able to do five years from now?
474
00:35:32.618 --> 00:35:34.219
And how do I anticipate that?
475
00:35:34.599 --> 00:35:50.367
And how does that change the number of people, the skills required, and what that organization... And that's across from physical labor, right, in the warehouse, driving trucks or whatever, all the way through to the C-suite.
476
00:35:50.387 --> 00:35:51.728
And same question about the future.
477
00:35:51.948 --> 00:35:55.410
And so what AI does is it makes a whole bunch of...
478
00:35:56.030 --> 00:36:00.111
new roles and functions potentially able to be transformed?
479
00:36:00.391 --> 00:36:04.853
And some of it might be how many people, but how would I operate differently?
480
00:36:06.413 --> 00:36:10.495
We talked about maybe a staff that has some AI in it.
481
00:36:10.535 --> 00:36:13.455
What is the question about spans and layers, right?
482
00:36:13.495 --> 00:36:16.897
Like, do I need to count AI as part of my spans and control?
483
00:36:16.977 --> 00:36:21.458
And if so, how many should I think about managing and all those sorts of things?
484
00:36:22.098 --> 00:36:25.819
And does AI actually allow me to increase my spans?
485
00:36:26.360 --> 00:36:36.203
Again, there are dystopic views about the way you can evaluate people, but maybe I can evaluate more people if my tests are good, the algorithm is aligned with the things I care about.
486
00:36:36.643 --> 00:36:39.804
And so those are the sorts, that's at least one set of questions.
487
00:36:40.425 --> 00:36:44.146
Let's look at what is my organization in the future in five years, and then how do I get there?
488
00:36:44.546 --> 00:36:49.468
And again, to the AI point, to the extent to which you need to do retraining or reskilling
489
00:36:49.848 --> 00:37:01.957
you know, that's another tool, not the only one, but another tool that could help you potentially get from here to there and carry forward some of your employees from today into tomorrow in a way that they're ready for that.
490
00:37:02.577 --> 00:37:04.619
And so that's, those are some of the questions that we see.
491
00:37:05.290 --> 00:37:11.315
You mentioned something interesting that I was hoping maybe you can unpack a little bit about leaders managing AI.
492
00:37:12.616 --> 00:37:15.139
And, you know, some people might be taken aback.
493
00:37:15.219 --> 00:37:16.680
You know, what does that mean?
494
00:37:17.340 --> 00:37:28.310
So can you explain a little bit about how and some people who are following, you know, the news or McKinsey, maybe they've seen some examples or visuals on that.
495
00:37:28.910 --> 00:37:35.937
But how would that work if I, as a leader, am managing a team of AI bots?
496
00:37:37.178 --> 00:37:40.601
Because people are having a hard time grasping, like, what does that even mean?
497
00:37:40.701 --> 00:37:42.503
What does that look like?
498
00:37:43.043 --> 00:37:44.024
How can that be?
499
00:37:44.685 --> 00:37:49.849
Like, what would be a scenario or an example of how that could become real?
500
00:37:49.969 --> 00:37:50.150
Yeah.
501
00:37:51.311 --> 00:37:57.253
So I use the most provocative version of that when I talked about, you know, what's your spending control and all those sorts of things.
502
00:37:57.273 --> 00:38:01.434
But like today's version is not quite like that, right?
503
00:38:01.854 --> 00:38:08.456
So take, for example, this class of product called Deep Research, which is a type of knowledge agent.
504
00:38:08.956 --> 00:38:20.200
And these systems are able to spend more time going out and, you know, they chat about, you expect the response within a couple seconds or something like that, but you can send off one of these...
505
00:38:21.140 --> 00:38:36.274
deep research or research agents to spend, you know, current versions are maybe up to 30 minutes, but, you know, people have contemplated and people experimented and like, what if you spend it, have it go off and work on a problem for a day and it comes back with a set of answers.
506
00:38:37.075 --> 00:38:41.879
Well, that's oftentimes what you might do with someone who reports to you, right?
507
00:38:41.960 --> 00:38:52.969
You might task them to go out and please investigate this market or, you know, go find out what all the comparable people are doing in terms of their HR policy around X and they'll come back and report to you.
508
00:38:53.590 --> 00:38:56.132
Well, now we have AI systems that can do that.
509
00:38:56.152 --> 00:38:58.595
These agents, they, you know, they do multi-step things.
510
00:38:58.615 --> 00:38:59.776
They're based on AI and.
511
00:39:00.616 --> 00:39:08.664
You know, the acting in the world is marginal, but creating a report is kind of like acting in the world, or it's a version of acting in the world.
512
00:39:09.264 --> 00:39:17.052
And so if you think about that, well, okay, then what are the sorts of things I need to do as a manager if I were going to go task somebody to do something?
513
00:39:17.072 --> 00:39:23.638
Well, I have to tell them in a sufficiently precise way that they go out and do the thing that I want to do.
514
00:39:23.718 --> 00:39:25.780
I also need to evaluate what comes back.
515
00:39:26.420 --> 00:39:28.181
And usually it's an iterative process, right?
516
00:39:28.301 --> 00:39:28.781
Oh, yeah, cool.
517
00:39:28.801 --> 00:39:29.862
Thanks for bringing me that.
518
00:39:30.042 --> 00:39:30.422
You know what?
519
00:39:30.702 --> 00:39:33.563
I think we should be able to do X, Y, Z. Yeah.
520
00:39:33.684 --> 00:39:47.370
And then that process, but instead of a person, how would you do that with an AI system, is some of the things that have to do with, you know, the quote-unquote direct supervision of an AI.
521
00:39:47.430 --> 00:39:49.951
But, you know, you might call it prompt engineering or something like that.
522
00:39:50.271 --> 00:39:53.693
And then eventually you potentially get to a place where, you know,
523
00:39:53.853 --> 00:39:58.314
Let's say you want to choose between this version of deep research and that version.
524
00:39:58.334 --> 00:40:03.495
Well, then you might be coming to something like evaluation and performance appraisal.
525
00:40:04.016 --> 00:40:04.916
We're not there yet.
526
00:40:05.796 --> 00:40:09.877
We're really not there yet because for performance appraisal, it's a year long type of thing.
527
00:40:09.897 --> 00:40:13.218
And that's not really what our AI types of systems are.
528
00:40:14.018 --> 00:40:14.558
are for.
529
00:40:14.618 --> 00:40:31.585
But, you know, again, to this point that Ethan Moloch made to me, oftentimes there are pretty close analogies to the way that we interact with human beings, to the way that we can effectively interact with AI, not being confused at all that the AI is a person.
530
00:40:33.686 --> 00:40:35.167
Yeah, that'll be interesting to see.
531
00:40:35.187 --> 00:40:38.188
I don't know if there are any examples out there yet,
532
00:40:44.654 --> 00:40:47.976
I mean, it can feel like that already, right?
533
00:40:48.077 --> 00:40:57.583
If I kick off a bunch of deep research, I'm thinking about things that are quite similar to when I'm managing people, like sequence.
534
00:40:58.804 --> 00:41:01.146
When are things going to come back?
535
00:41:01.266 --> 00:41:02.067
Just, you know...
536
00:41:02.887 --> 00:41:16.600
All of that feels... Knowing how to leverage these tools is also... And I saw there was even like job postings for prompt engineers that were paying like a quarter million dollars.
537
00:41:17.501 --> 00:41:21.545
So it's one thing to have the tool, but it's another thing to actually know how to use the tool.
538
00:41:21.665 --> 00:41:22.586
And so I wonder...
539
00:41:24.686 --> 00:41:33.972
Is there a big gap between the capabilities that these AI tools have today versus the knowledge that employees have of how to use these AI tools?
540
00:41:34.653 --> 00:41:44.240
Because if you give ChatGPT or a similar system to a lot of employees, they might not know what to do with it.
541
00:41:44.360 --> 00:41:45.781
What questions do I ask?
542
00:41:45.821 --> 00:41:48.102
How do I frame the prompts the right way?
543
00:41:48.123 --> 00:41:48.223
Yeah.
544
00:41:49.332 --> 00:42:08.587
Uh, you know, a lot of people don't know that you also have to double check a lot of things like there, there have been times I even tested it with some like McKinsey reports where I would put it into a chat GPT and I'll say, you know, pull out whatever interesting insights or I'm writing something on AI, give me some interesting insights and it'll cite a McKinsey report.
545
00:42:08.847 --> 00:42:09.908
And then I'll find that report.
546
00:42:10.188 --> 00:42:14.512
I'll pull it up and I'll look for that quote or that citation or that statistic.
547
00:42:14.952 --> 00:42:15.513
And I don't find it.
548
00:42:16.478 --> 00:42:19.119
And then I have to go back to the chat GPT and I have to say, hey, wait a minute.
549
00:42:19.479 --> 00:42:22.699
You told me that there's this stat from the report.
550
00:42:23.000 --> 00:42:23.540
I don't see it.
551
00:42:24.320 --> 00:42:29.361
And then after going back and forth with the bot, like I'm arguing with a person, it'll finally come back to me and say, oh, you're right.
552
00:42:30.141 --> 00:42:38.123
I extrapolated based on that report of what they could have said if they would have said this based on the way that they're writing things.
553
00:42:39.303 --> 00:42:44.665
But that's not the same thing as like a stat that you actually said, right?
554
00:42:44.825 --> 00:42:45.965
It's making stuff up.
555
00:42:46.933 --> 00:42:51.795
So you also have to know how to push back on these tools, just like you would with a person.
556
00:42:51.895 --> 00:42:52.635
Are you sure?
557
00:42:53.075 --> 00:42:54.216
Did you double check?
558
00:42:54.676 --> 00:42:55.896
Where are your sources?
559
00:42:55.956 --> 00:42:57.457
Can you cite the sources?
560
00:42:58.437 --> 00:43:04.920
So are you seeing a gap between capability versus our knowledge as humans of how to use these tools?
561
00:43:06.023 --> 00:43:06.623
The answer is yes.
562
00:43:07.203 --> 00:43:09.884
That gap does close with experience and practice.
563
00:43:10.845 --> 00:43:12.485
And that gap can close also with training.
564
00:43:13.246 --> 00:43:16.667
But to your point, you know, the hallucination problem is real.
565
00:43:16.707 --> 00:43:19.968
It's, you know, the technologists are working really hard on it.
566
00:43:20.008 --> 00:43:21.029
The hallucination meaning...
567
00:43:21.969 --> 00:43:26.372
coming up with wrong stuff, inaccuracy or made up stuff, as you said, that, that is actually real.
568
00:43:28.294 --> 00:43:40.302
But, you know, I often describe at least the current versions of these as we sit today as being, you know, maybe like a summer intern, summer interns that can do real work, right.
569
00:43:40.322 --> 00:43:47.007
Then, you know, there are some places that, but the people we bring in on the summers, we make, you know, they work, they work really hard.
570
00:43:47.973 --> 00:43:52.854
But we wouldn't take their output and just forward it immediately to a CEO.
571
00:43:54.175 --> 00:44:00.917
We have a whole set of ways that we check things, that we make things better.
572
00:44:01.777 --> 00:44:13.860
And I think, again, to this analogy, again, not confusing machines with people, but to your point, we have ways of testing whether or not the output is good.
573
00:44:14.301 --> 00:44:15.681
We can help make it better.
574
00:44:16.721 --> 00:44:17.422
checking things.
575
00:44:18.442 --> 00:44:29.508
And so I think that type of analogy is really useful as we think about how can we use these systems, which are imperfect, but productively.
576
00:44:31.349 --> 00:44:37.072
Yeah, that is, and I think a lot of organizations are in the process of going through training.
577
00:44:39.355 --> 00:44:42.536
you know, some mandatory, some optional training.
578
00:44:43.156 --> 00:44:50.218
But I wonder, I guess it's through trial and error because even when I learned how to use ChatGPT, I didn't take a training per se.
579
00:44:50.238 --> 00:44:56.699
In fact, I remember when I first had access to it, I would just put in like these stupid questions in there and just see what it would give me.
580
00:44:56.739 --> 00:45:00.820
And then over time, you start to say, hey, wait a minute, you're actually pretty...
581
00:45:01.778 --> 00:45:02.600
pretty capable here.
582
00:45:03.080 --> 00:45:04.744
You can actually do some stuff.
583
00:45:05.705 --> 00:45:12.738
So I guess it's, um, don't be scared to tinker and play around with it and don't, uh, you know, don't run away from it.
584
00:45:12.818 --> 00:45:13.980
Maybe it's kind of the lesson there.
585
00:45:14.766 --> 00:45:21.112
And I'll tell you that this is one of the remarkable things I think about, you know, I'm a former CIO.
586
00:45:21.332 --> 00:45:30.660
And a lot of times the enterprise technologies we had, you know, we really did have to train people how to use them because they were difficult.
587
00:45:30.720 --> 00:45:32.842
And in school, you needed the instruction manual.
588
00:45:33.262 --> 00:45:36.605
The amazing thing about consumer products, right?
589
00:45:36.685 --> 00:45:40.129
Like, I presume that you weren't trained on using Instagram, right?
590
00:45:40.809 --> 00:46:05.156
that and and using it you use it why because it's so compelling and experience you know and we hire all these people and they come in and you say why is my enterprise stuff so crappy right like you know your aspiration as someone who creates enterprise software should be something that's so great that number one you don't need an instruction manual you don't need formal training number two it's so great that you're
591
00:46:05.736 --> 00:46:07.637
people use it because they want to use it.
592
00:46:08.297 --> 00:46:12.918
And so I think a lot of these tools are starting to feel a little more like that.
593
00:46:12.958 --> 00:46:14.018
Like, why aren't you using this?
594
00:46:14.058 --> 00:46:14.738
It's pretty cool.
595
00:46:14.758 --> 00:46:17.499
And you say, well, it gives me hallucinations.
596
00:46:18.179 --> 00:46:20.560
Well, we should go make sure we check it before we use it.
597
00:46:20.580 --> 00:46:30.342
But it's still really, and that's the kind of dialogue that I think we potentially could see that, again, yes, formal training is helpful.
598
00:46:31.283 --> 00:46:37.910
Um, but a lot of times I think what we do is we gain experience a lot like management itself, right?
599
00:46:37.930 --> 00:46:39.852
I mean, absolutely.
600
00:46:40.052 --> 00:46:43.155
Going to management training is incredibly useful and important.
601
00:46:43.976 --> 00:46:48.381
And you learn a lot about managing people by managing people.
602
00:46:48.962 --> 00:46:49.122
Hmm.
603
00:46:50.187 --> 00:46:50.887
Yeah.
604
00:46:51.307 --> 00:46:56.269
In my first book or one of my earlier books, the employee experience advantage, I called it consumer grade technology.
605
00:46:56.650 --> 00:46:56.810
Yeah.
606
00:46:57.010 --> 00:46:57.170
Right.
607
00:46:57.210 --> 00:47:05.173
The, the tools you use in your person, in your work environment should be so beautiful, useful and valuable that you would want to use something in your personal life if something like that existed.
608
00:47:06.033 --> 00:47:13.836
So yeah, that's, uh, I remember that the big debates we had around those collaboration softwares, way back in the day.
609
00:47:14.997 --> 00:47:15.177
Um,
610
00:47:15.725 --> 00:47:16.526
Maybe we could wrap up.
611
00:47:16.566 --> 00:47:22.350
I know we only have a couple of minutes left to look at the state of AI today and where it's going.
612
00:47:22.530 --> 00:47:28.215
Because for a lot of people, I think they're still trying to understand just how good is AI now?
613
00:47:29.156 --> 00:47:31.578
Like, what can it actually do?
614
00:47:31.718 --> 00:47:33.259
How much information does it have?
615
00:47:34.040 --> 00:47:44.568
So I don't know if you're able to paint a picture on the capability of AI today and also what you think the capability of AI is gonna be in three to five years.
616
00:47:46.880 --> 00:47:49.581
It's moving so fast at some level, right?
617
00:47:49.601 --> 00:47:52.042
Which is that literally every day there's another announcement.
618
00:47:52.122 --> 00:47:53.423
Again, I live in San Francisco, right?
619
00:47:53.443 --> 00:47:59.225
So I mean, people are, you know, there's a launch of something, you know, multiple times a day.
620
00:47:59.285 --> 00:48:01.867
And so things are moving very, very quickly.
621
00:48:05.027 --> 00:48:07.890
So it's hard for me to just sort of synthesize it all up.
622
00:48:07.990 --> 00:48:13.994
I mean, I think, again, if you have the experience of using these technologies, what's remarkable is, yes, you can have a conversation with it.
623
00:48:14.014 --> 00:48:17.057
And you have a conversation with it that is people-ish.
624
00:48:17.617 --> 00:48:21.580
They are becoming more multimodal in the sense that it's not just language.
625
00:48:21.660 --> 00:48:30.027
But, you know, you can have a conversation, have your camera up and you can have a conversation about something, quote, unquote, it's looking at and you're looking at at the same time.
626
00:48:30.767 --> 00:48:44.074
And increasingly, you know, the agents thing is you can send these things off to go and do some work for more of an extended period of time rather than just, you know, responding to you in real time or near real time.
627
00:48:44.114 --> 00:48:45.974
You can send it off to do some work.
628
00:48:47.235 --> 00:48:54.598
I think that's, and right, I mean, the whole web is accessible to these systems now, for instance.
629
00:48:54.638 --> 00:49:03.161
And so I think maybe, you know, just that's maybe sort of a state of the art today as a synthesis.
630
00:49:04.401 --> 00:49:05.522
And where are things going?
631
00:49:06.683 --> 00:49:12.370
The recent ChatGPT update, it does great infographics, great visuals, the text.
632
00:49:13.432 --> 00:49:13.992
Not perfect yet.
633
00:49:14.012 --> 00:49:17.898
It still makes some mistakes here and there, but that even got better.
634
00:49:18.018 --> 00:49:22.544
So it's definitely making leaps and progress in terms of leaps and bounds.
635
00:49:23.976 --> 00:49:30.638
You know, and one other thing that we're studying, too, there again, there for a long time, AI was stuff you didn't see.
636
00:49:31.158 --> 00:49:31.318
Right.
637
00:49:31.358 --> 00:49:34.459
I mean, it was like improving forecasting and they're really valuable stuff.
638
00:49:34.599 --> 00:49:34.859
Yeah.
639
00:49:34.979 --> 00:49:36.720
You know, improving next product to buy.
640
00:49:37.140 --> 00:49:43.822
There's a version of generative AI, which is also stuff you didn't see, because when we talk about large language models, it can produce language.
641
00:49:44.622 --> 00:49:48.424
But drug companies are using foundation models in order to create new drug candidates.
642
00:49:48.884 --> 00:49:57.548
And again, if you think about the potential for this to accelerate innovation in all kinds of fields, you know, curing disease, it's incredible what's possible.
643
00:49:57.588 --> 00:50:00.650
And so I think that's another place where we'll see more and more capability.
644
00:50:00.670 --> 00:50:03.511
But we won't see it, you know, it won't be right in front of us.
645
00:50:04.172 --> 00:50:11.295
It might show up as, you know, life extension drugs and all these kinds of things, which, you know, gosh, that would be great if we could have it.
646
00:50:11.809 --> 00:50:16.130
Yeah, I wonder if, you know, it reminds me like years ago, everybody would talk about IBM's Watson.
647
00:50:17.411 --> 00:50:19.331
But you don't hear anybody talk about it anymore.
648
00:50:19.731 --> 00:50:22.932
And it was this big promise to do like all these wonderful and amazing things.
649
00:50:22.972 --> 00:50:27.833
And then today, I mean, I'm assuming it's still around, but I don't know what's going on with it.
650
00:50:28.573 --> 00:50:29.254
And so...
651
00:50:31.790 --> 00:50:38.835
I can't imagine this will go down that same path as being so much hype and then very little action later on down the road.
652
00:50:39.395 --> 00:50:43.057
But that's certainly what happened to, I think, those early AI models, right?
653
00:50:43.077 --> 00:50:44.919
There was so much hype and promise behind it.
654
00:50:45.059 --> 00:50:48.761
And then fizzled out, disappeared.
655
00:50:49.081 --> 00:50:50.082
Nobody talks about it anymore.
656
00:50:51.051 --> 00:50:52.672
But there are two versions of disappearing.
657
00:50:52.952 --> 00:50:56.935
There's one version of disappearing is it's disappointing technology that doesn't have impact.
658
00:50:57.296 --> 00:51:05.322
There's another version of it disappearing, which I think happens all the time with technology, is we no longer, it's no longer remarkable, but it underpins a ton of value.
659
00:51:05.762 --> 00:51:10.105
And I think that's a lot of what will happen to AI and has actually happened to AI previously, right?
660
00:51:10.125 --> 00:51:13.228
I mean, there used to be AI, now we just call it product configurator, right?
661
00:51:13.288 --> 00:51:18.091
I mean, you know, there are all these things, it's suddenly not remarkable anymore.
662
00:51:18.592 --> 00:51:19.993
And I think that will happen a lot.
663
00:51:20.133 --> 00:51:21.114
It just becomes ambient.
664
00:51:21.634 --> 00:51:21.814
Yeah.
665
00:51:23.035 --> 00:51:23.775
All right.
666
00:51:23.795 --> 00:51:24.876
And then where do you think we're going?
667
00:51:25.536 --> 00:51:31.400
So five years from now, are we going to have humanoid bots?
668
00:51:32.060 --> 00:51:34.442
Are you going to have an AI co-worker?
669
00:51:35.362 --> 00:51:43.007
Are we going to have artificial general intelligence AI companies where like the CEOs, AI bot?
670
00:51:43.567 --> 00:51:46.089
Like where is all of this madness headed?
671
00:51:46.869 --> 00:51:49.832
We're going to have a whole other discussion about humanoid robots.
672
00:51:49.852 --> 00:51:51.894
You know, super interesting one as well.
673
00:51:51.934 --> 00:51:54.637
I mean, you know, hard tech or physical tech is hard, right?
674
00:51:54.657 --> 00:51:58.281
I mean, these things, you see the videos, but they're only a few minutes long.
675
00:51:58.621 --> 00:51:59.782
These systems typically.
676
00:51:59.962 --> 00:52:03.346
From DARPA, the things that Elon Musk is building, it creeps me out.
677
00:52:03.873 --> 00:52:10.096
Yeah, but they only have enough power to go for maybe four hours doing something light duty, and then they have to recharge for another two hours, right?
678
00:52:10.116 --> 00:52:12.416
That's a really long coffee break in the middle of your shift, right?
679
00:52:12.436 --> 00:52:16.358
So there's all these hidden things about physical technology, which are true.
680
00:52:16.898 --> 00:52:22.460
But in terms of the AI, the cognitive technologies, it will get more and more powerful over time.
681
00:52:23.441 --> 00:52:30.504
And it is going to be something that, again, the echo of history, what's happened is...
682
00:52:31.525 --> 00:52:34.348
You know, it's quite rare that an occupation disappears.
683
00:52:34.968 --> 00:52:39.112
What mostly happens is the activities you do change over time.
684
00:52:39.132 --> 00:52:41.515
And again, I'm just going to use an overall analogy.
685
00:52:42.015 --> 00:52:48.802
But oftentimes what happens is, you know, we talked about this a little bit, you know, rather than doing the work, you become the manager of the work.
686
00:52:49.262 --> 00:52:49.422
Right.
687
00:52:49.442 --> 00:52:50.684
And so this has happened in factories.
688
00:52:50.764 --> 00:52:50.984
Right.
689
00:52:51.024 --> 00:52:53.006
Rather than, you know, me turning the bolt or something like that.
690
00:52:53.326 --> 00:52:57.547
Now I'm managing a set of robots that do that kind of thing, or a set of machines that do that kind of thing.
691
00:52:58.128 --> 00:53:00.108
I think that's going to happen to our cognitive work as well.
692
00:53:00.528 --> 00:53:07.991
And so we will have more of these machines doing some of our cognitive work, but people are now going to level up, saying like, how do I manage
693
00:53:08.615 --> 00:53:12.658
all this work and work includes like trying to find mistakes, right?
694
00:53:13.058 --> 00:53:14.259
But also improve things.
695
00:53:14.279 --> 00:53:16.240
How do I have a dialogue that allows us to happen?
696
00:53:16.280 --> 00:53:20.163
How do I make the whole entire process more efficient and effective?
697
00:53:20.203 --> 00:53:23.425
I think that's one of the things that's going to, to, to level up.
698
00:53:23.505 --> 00:53:26.588
I think the other question that comes to mind is, okay, okay.
699
00:53:27.188 --> 00:53:31.711
You know, what are the skills and occupations that will be in demand going forward?
700
00:53:31.731 --> 00:53:34.533
Because again, if, if, you know, cognition is cheap, right?
701
00:53:34.913 --> 00:53:37.156
And the physical stuff is hard, as I just made the argument for.
702
00:53:37.176 --> 00:53:39.038
Well, maybe that becomes more valuable.
703
00:53:39.078 --> 00:53:40.520
But then, you know, how do we pay for it?
704
00:53:40.560 --> 00:53:41.301
What do we pay for it?
705
00:53:41.761 --> 00:53:46.026
You know, these are the sorts of questions that I think are really important as a society going forward.
706
00:53:46.046 --> 00:53:52.013
But I think, you know, just in terms of the workplace, we're going to see more and more of these powerful cognitive machines.
707
00:53:53.290 --> 00:53:56.271
Yeah, the next few years will be very, very interesting indeed.
708
00:53:57.471 --> 00:54:00.352
Michael, where can people go to learn more about you?
709
00:54:00.732 --> 00:54:04.593
I mean, you have a lot of really great reports and things like that that you publish on McKinsey.
710
00:54:05.153 --> 00:54:07.954
What do you do, like a couple every year, like once a quarter or something?
711
00:54:08.094 --> 00:54:09.234
You have a lot of really cool things.
712
00:54:09.314 --> 00:54:13.435
Yeah, I mean, the big ones come less frequently, the small ones more frequently.
713
00:54:13.495 --> 00:54:16.456
But McKinsey.com, it's free.
714
00:54:16.556 --> 00:54:17.756
So feel free to take a look.
715
00:54:18.216 --> 00:54:20.617
And are you publishing on LinkedIn yourself frequently?
716
00:54:21.615 --> 00:54:24.696
Um, I point to the other stuff that I published from LinkedIn.
717
00:54:25.216 --> 00:54:25.516
Okay.
718
00:54:25.776 --> 00:54:25.996
All right.
719
00:54:26.016 --> 00:54:31.758
So people can just type, uh, type in Dr. Michael Chewy McKenzie, and they can find your, uh, all your stuff on there.
720
00:54:32.138 --> 00:54:32.298
Yep.
721
00:54:33.219 --> 00:54:33.479
All right.
722
00:54:33.539 --> 00:54:35.639
Well, thank you so much for taking time out of your day.
723
00:54:35.659 --> 00:54:36.340
I really appreciate it.
724
00:54:36.360 --> 00:54:42.742
I felt like we could have kept talking for another hour, but, uh, you know, I'll have to have you back, uh, back on the show to talk about the robots next time.
725
00:54:43.522 --> 00:54:44.222
Thanks for having me on.
726
00:54:45.937 --> 00:54:49.598
Thanks for tuning into today's episode of Future Ready Leadership.
727
00:54:50.039 --> 00:54:56.921
If you want to get access to more content on how you can become future ready, make sure to subscribe to the newsletter.
728
00:54:57.321 --> 00:55:02.643
It's thefutureorganization.com forward slash newsletter.
729
00:55:03.043 --> 00:55:07.365
And if you enjoyed this episode, please send it to somebody who you would like to see become a better leader.
730
00:55:07.805 --> 00:55:10.146
Thanks for listening or watching, and I will see you next time.