0:00

it's Mitch with just a little update on

0:02

my new venture, Thinkers One. As

0:04

a reminder, we've built a way for

0:06

organizations of all shapes and sizes

0:08

to buy bite -sized and personalized thought

0:10

leadership video experiences from the best thinkers

0:13

in the world. So what's changed?

0:15

Well, we've realized that we need to

0:17

make these thought leaders much more

0:19

accessible. So when you head over to thinkersone

0:21

.com, you will see that the pricing

0:23

is very reasonable. Go and check it

0:25

out. And all of these experiences

0:27

are at least 15 minutes of pure,

0:29

brainy goodness. Now you

0:31

can bring world -class authors, speakers, academics,

0:33

captains of industry and more into your

0:36

everyday meetings or gatherings. It's

0:38

really simple to use and it's

0:40

such a unique way for business leaders

0:42

like you to buy completely personalized

0:44

video content. You can book these

0:46

thought leaders to pop in live at your

0:48

next meeting or have a thinker record

0:50

a presentation just for your team or even

0:52

as a great gift for clients. I'm

0:54

really proud of the lineup of thinkers that

0:56

we have. We've got great minds like

0:58

Tom Peters, Rita McGrath,

1:00

Michael Bunga -Stanye, Whitney Johnson,

1:03

Mark Bowden, Dory Clark, Jay Baer, and

1:05

so many more. So if you're looking

1:07

to add insights, energy, and big smarts

1:09

to your next meeting, corporate event, an

1:11

offsite, a lunch and learn, even as

1:14

a gift for that client, I know

1:16

that you're going to love this. I invite

1:18

you to head over to thinkersone .com and check

1:20

it out. Now, back to the show. Well,

1:22

hey there and welcome to

1:24

episode number 972 of Six

1:26

Pixels of Separation, the Thinker's

1:28

One podcast. My name

1:31

is Mitch Joel. It's Sunday, February

1:33

the 23rd, 2025. Let's

1:35

get on with the show. So

1:49

who are you and what do

1:51

you do? Hello, I'm Dr. Tom Chapfield

1:53

and I'm a philosopher of technology. I

1:55

write books designed on my courses and

1:57

think about what it means to use

1:59

technology well. That's great. You got

2:02

a great new book out called Wise Animals that

2:04

I totally loved. So thank you for writing that

2:06

one. Thank you. My great pleasure. Now,

2:08

I would love to be a philosopher in tech. Tell

2:10

me what I got to do. How does this work?

2:13

Well, It is, of course, a bit

2:15

of hand -waving. But for me, my background

2:17

is literature and philosophy, but

2:19

being a geek. So I did a doctorate

2:21

at Oxford in literature and philosophy, but

2:23

I was obsessed with the fact that all

2:26

of the stuff I was writing and

2:28

thinking about, what it means to live well,

2:30

to relate to others, to think well,

2:32

and so on, was bound up

2:34

with technology. And I felt there

2:36

was a real gap here and a

2:38

real need, a need to try and be

2:40

more thoughtful about the texture of the

2:42

technologized world. And so I started working

2:44

and writing books around things like video

2:46

games around online politics around kind of

2:48

studying a digital age. And I guess

2:50

I followed it from there. So I

2:52

call myself a philosopher because people don't

2:54

quite know what it is and then

2:56

I get to explain it. But what

2:58

I love to do is try and

3:01

get people to think twice about the

3:03

technologies they use, how they work with

3:05

and through technology and businesses, organisations. It's

3:08

helping people be more thoughtful. individual

3:10

institution and kind of make the

3:12

most of the human in a digital

3:14

age. Is it more about how

3:16

you editorialize on what you're seeing

3:18

in terms of research? Is that

3:20

what it would be? For

3:23

me, it's that. I mean, I'm

3:25

trying to come up with actionable

3:27

insights, really. So I chair several

3:29

organizations. I work at advisory board.

3:31

I have my own company. I

3:33

try and take research

3:36

around cognition, cognitive bias,

3:38

behavior. ethics. And

3:40

I try and help people understand how it can

3:42

be applied. So in some ways, although I love

3:44

to write, and I've written a

3:47

dozen books now, I think

3:49

it's my kind of core activity,

3:51

I'm really preoccupied with making

3:53

it actionable and powerful. And

3:55

in particular in boardrooms, for example, what

3:57

I often find myself doing is trying

3:59

to come up with if you'd like kind of thinking tools

4:01

and strategies in a common language. So with

4:03

AI, you often have organizations not

4:05

knowing how to talk and think about

4:07

what it is. not having the right

4:09

ways of talking about what these powerful

4:11

technologies are, how they can use their

4:13

own skills, where their skills gaps are.

4:15

And you end up with people saying,

4:18

well, you know, it's up to the

4:20

tech people, it's up to the CTO,

4:22

it's up to this product. They're not

4:24

making confident informed values led decisions. And

4:26

so this hopefully is where I come in.

4:28

So I'm a little bit obsessed with

4:30

not just sort of talking the talk, but

4:33

trying to make that talk

4:35

into kind of techniques and

4:37

practical interventions and ways of

4:39

doing as well as thinking.

4:42

Are those surprising conversations, meaning

4:44

are these executives hearing what

4:46

you're saying and then thinking

4:48

and pausing and reflecting on this idea

4:50

that that's true? Why do I allow

4:52

technology to sit within the IT department

4:54

or marketing? Or do you

4:56

find them still in the

4:58

mindset of it's a tool? So

5:01

there's often a self -selection thing here because, of course,

5:03

if someone is going to pay me to come in

5:05

and work with them, They're probably interested in the

5:07

kind of things I have to do. But

5:09

you certainly can see that there are sometimes

5:11

people within an organization who want to be

5:13

more thoughtful and make a change in people

5:15

who don't. And the question is what it

5:17

means to get everybody on the same page

5:19

or as far as possible. And

5:21

I think one of the things you

5:23

can do here is really focus on

5:25

people's day -to -day experiences. So

5:27

one of the things I always talk about in a

5:30

very simple way is, well, technology

5:32

is not just a tool. in

5:34

terms of being a neutral thing that you

5:36

just use or don't use. Technology

5:38

has certain biases, certain

5:40

assumptions baked into

5:42

it. Tech of technology is simple

5:44

as email. And then

5:46

ask yourself a deceptively silly question.

5:49

What does email want me to do?

5:51

What behaviors is email pushing me

5:53

towards? And you see when

5:55

you talk to people that they notice that

5:57

with email, what email wants you to

5:59

do is send more

6:01

email. It's like a to

6:04

do list written for you by other

6:06

people because it's cost -free and time

6:08

-free to send email messages and in

6:10

most modern workplaces you send email

6:12

to prove that you're working, that you're

6:14

attending that you're checking in, you

6:16

copy people in, you send attachments, you

6:18

send updates when you're away from

6:20

your desk, you have it

6:22

automatically email people to tell

6:25

them that you'll come back and send

6:27

them more email from a certain perspective. 50

6:29

% of what many highly

6:31

accomplished people do is effectively

6:33

work as email inbox emptiers. But

6:36

of course, by emptying their inboxes,

6:38

they fill up the inboxes of everybody

6:41

else in their organization and beyond. And

6:43

once you surface this, you can say, well,

6:45

actually, how is the tool

6:48

working for you? What would

6:50

an informed negotiation between your needs

6:52

and priorities? And if you

6:54

like the affordances, the kind of

6:56

structural predilections or biases the tool look

6:58

like, a lot of organizations of course

7:00

have done this thinking and are saying,

7:02

well, we're going to have communication protocols.

7:05

We're going to insist that people use

7:07

Slack or shared documents rather than

7:09

email for certain things. We're going to

7:11

insist that famously at an organization

7:13

like Amazon, if we're sitting down

7:15

to a meeting face to

7:17

face, we are preparing and discussing

7:20

a one pager or a

7:22

two pager. a reasoned argument or

7:24

debate. We're accessing data separately.

7:26

We're not sitting through an endless

7:28

PowerPoint. So once you get down

7:30

to these discussions, you can then do

7:32

the really interesting thing, which is say, well, let's

7:35

first of all just talk about

7:37

what you're doing, about what the

7:39

technologies in your working life do, and

7:41

then about what behaviors they push

7:43

you towards, what assumptions or

7:45

limitations they may encode. and

7:48

what it means to push back against these assumptions.

7:50

So I love the practical psychological stuff,

7:52

and I won't bore you yet with

7:54

examples, but there's a lot more where

7:57

that came from. Do you feel that

7:59

part of this is a cycle of

8:01

evolution? And what I mean by

8:03

that, Tom, is as you were

8:05

talking like this, I was reflecting on

8:07

the fact that I wonder if

8:09

we push this 20 years into the

8:11

future, that if you had simply

8:13

digital natives who were engaged in this,

8:15

that This wouldn't even be the

8:17

thought. It would almost be like a fish

8:19

talking about water. Do we

8:21

feel like these are problems because we

8:23

have this weird cycle where it's

8:25

mostly Gen X or isn't it hell,

8:28

who are in these sea suites

8:30

versus if somebody was a digital native, it

8:32

would just float through and they wouldn't be

8:34

thinking about life before type of thing, which

8:36

gives us the thought of perhaps we're not

8:39

doing it right. So I would flip

8:41

this around and say one of the

8:43

great assets of a kind of a

8:45

workplace with people from different generations and

8:47

also a world in which you've

8:49

got someone like me who can remember

8:51

a time before smartphones and ubiquitous

8:54

internet. We can then

8:56

try and collectively capture the best

8:58

of the old and the best of the

9:00

new. I spend a lot of time in

9:02

universities and schools and colleges talking to 15,

9:04

16, 17, 18, 19 year

9:06

olds, so -called digital natives. I

9:08

think the world is a bit of

9:10

a myth. In that again and again I

9:13

find that people are very very good

9:15

at certain things very fluent but also they

9:17

don't know what they don't know they

9:19

often may not realize that the young age

9:21

for example that resources like Wikipedia and

9:23

so on are user generated you look at

9:25

the edit histories find out where they're

9:27

from people certainly don't realize with

9:30

AI how a large language

9:32

model works that often it's

9:34

outlets are probabilistic. rather than

9:36

factual. And again, people don't

9:38

realise simpler things like how they're

9:40

being tracked, like how they might be

9:42

being surveilled, or indeed how to

9:44

make their computers safe, or the benefits

9:46

of being offline, the benefits

9:49

of certain emotional, empathetic soft

9:51

skills. I spend a lot

9:53

of time working with 17,

9:55

18, 19 year olds about

9:57

pausing, reflecting, thinking

9:59

twice. So for

10:01

me, the prize is that

10:03

exactly Like

10:06

you said you don't want to have a

10:08

world of fish you don't realize there

10:10

in water because of course the difference

10:12

between us and fish is

10:14

that we can reflect upon our environment

10:16

and the environment through which we

10:18

swim within which we live a human

10:20

made and human maintained. You've

10:22

heard I'm sure of the idea

10:24

of dark patterns of set ups in

10:26

websites and interfaces which are

10:29

covertly manipulative. where you're automatically opted

10:31

in to something where you

10:33

can't cancel your subscription, where your

10:35

data is harvested, where the

10:37

real options are hidden from you.

10:40

The problem is that there's not just one kind

10:42

of water out there. There's lots of different

10:44

kind of lakes and oceans that are being run

10:46

by companies for certain reasons. So

10:49

for me, I love going

10:51

to people and doing two

10:53

things really, saying, first of all,

10:55

let's celebrate your knowledge. And

10:57

let's celebrate your knowledge by getting you to share

11:00

with each other your best tricks and tips.

11:02

If you're 20, what would you tell your 16

11:04

year old self? If you're 16, what would

11:06

you tell your 12 year old self in terms

11:08

of communications protocols in terms of safety? And

11:10

a lot of people say, okay, here's the simplest

11:12

thing I do. I'd say, wait a minute

11:14

before you send that email. My

11:16

favorite bit of, is it where

11:19

tech hackery I have is very

11:21

simply a 30 second delay on

11:23

my email when the 30 seconds after I've

11:25

sent any message I can read it and take

11:27

it back. And my goodness, I

11:30

take back a lot of those messages because I realized

11:32

that I'm passive aggressive, or that I haven't sent the

11:34

attachment, or that I'm boring, or I didn't need to

11:36

be sent. And then the second thing

11:38

is to try and, by comparing people's

11:40

knowledge and experience, find

11:43

out what their unknown unknowns are.

11:45

Someone is 40, someone is 50,

11:47

someone is 30, they're going to

11:49

have different gaps in their knowledge.

11:51

And the big thing with modern

11:53

organizations, with complex systems, is covering

11:55

the gaps, is pooling your

11:57

knowledge constructively and then

11:59

is coming up with the best possible strategy. And

12:02

that really does mean, I think, not

12:04

taking things for granted. Last simple

12:06

example, for most people, again,

12:08

a very powerful intervention in terms of

12:11

people's relationship with technology can be an

12:13

audit of apps on their phone. It

12:15

can be simply getting people to pause And

12:17

because these things have built up, there's a

12:19

lot of momentum. It's like a QWERTY keyboard.

12:21

It's not a QWERTY keyboard because it's the

12:23

best design in the world. It's because it's

12:26

always been like that because of the peculiarities

12:28

of the mechanics of typewriters in the century.

12:31

So do what you can, audit your

12:33

apps, look at your day, look

12:35

where your time's going, reflect, improve, iterate,

12:37

do this together. That for

12:39

me is the recipe. Yeah, the evolution

12:42

that I was talking about, my experience today

12:44

as a first time, and I thought it

12:46

was very fortuitous that I was sitting down to

12:48

speak with you about this, which was new iOS

12:50

comes out, I updated, it

12:52

includes a version of chat GBT,

12:54

I include it. And

12:56

suddenly I found this feature where I go

12:58

to hit reply to an innocuous email,

13:01

is this time okay for this meeting? And

13:03

instead of doing that reply of thank you

13:05

so much for reaching out and I look

13:07

forward to the meeting, there was the little

13:09

AI button and I press that. And I

13:11

noticed that it formulated the response for me.

13:15

And I reviewed it and thought, 80

13:17

% right. Thanks so much for reaching out. Then there's

13:19

some information on it that I was, you know,

13:21

don't need that. And much like you,

13:23

I will take that buffer time before

13:25

I send the response. And I realized that

13:27

I don't need that buffer time anymore

13:30

because just by seeing its response and thinking

13:32

about it and thinking what would be the correct

13:34

way to correct it, it actually gave

13:36

me more insights. And

13:38

more thought about what I should

13:40

say to avoid being passive -aggressive or

13:42

typically I would be sarcastic. So

13:44

that's a great example of, I

13:46

guess, what Ethan Molyk around others

13:48

has pointed out as the result

13:50

of a kind of a new

13:52

suffusion of intelligence. And

13:54

in terms of the human skills, I

13:56

think you're absolutely right that suddenly the skill

13:58

you're using is the

14:00

skill of reflection and editing.

14:03

and you're not being your own typist.

14:05

And that can be a very good thing.

14:07

And previously, of course, we had to type

14:09

all our messages. It just goes without saying. Now

14:12

we don't. So the question becomes, which

14:15

messages should we type and why?

14:17

And which messages should we just

14:19

review? Which should we outsource and

14:21

delegate? I'd like to be more pretentious

14:23

with my thought and think. Please. That

14:26

seems to me almost, let's call it

14:28

level one of how we react to it.

14:30

My level two or level three thinking

14:32

on this is Just hold

14:34

on a second. Did I

14:36

just enter a new phase

14:38

where my AI is talking

14:40

to your AI essentially and

14:42

then more philosophically than what am I?

14:45

Yeah. And of course, absolutely. And I

14:47

think you are an intensified version

14:49

of what you've always been because to

14:51

be human is to be a technologized

14:53

animal. I am already a different

14:55

person with my phone than without it.

14:57

I'm different to my ancient ancestors because I'm

15:00

literal. because I read, I

15:02

write, I use screens, I communicate

15:04

with you right now at the speed

15:06

of light across the world from a little

15:08

office in my garden. So this

15:10

is an intensification of something that has

15:12

always been true that as human

15:14

beings we have this astonishing ability to

15:16

change ourselves through

15:18

technology. Our minds are

15:21

literally extended through technologies

15:23

and we acquire this

15:25

after birth due to

15:27

this extraordinary neuroplasticity. And

15:29

the point I'd make is that this

15:31

both isn't isn't new so we are this

15:33

is profoundly new but for example

15:35

the analogy of digital photography is interesting

15:37

takes photography from being something where film

15:39

is costly and expensive and you take

15:41

a few photos and it's for a

15:43

smaller number of people gradually becomes a

15:45

mass art form and now digital

15:47

photography. Your camera is taking

15:50

beautiful shirt photos of anything you want

15:52

kind of infinitely and suddenly as you say

15:54

you have this at your fingertips it's

15:56

almost like a part of your hand and

15:58

then you're kind of the conductor. Of

16:00

the orchestra you're the operator of the

16:02

machine you're the manager of this

16:04

kind of little cottage industry you're managing

16:06

a whole bunch of bits of

16:08

software and hardware that are kind of

16:10

running the identity that is you. You're

16:13

in a kind of director mode and

16:15

you're making decisions like okay well

16:17

which photo shall I show who shall

16:19

I send them to which how shall

16:21

I market them advanced and talk

16:23

about them can i turn them into data can

16:25

they be useful data that i'll feed back

16:28

into the machine. I write with AI a lot,

16:30

not my books, but I write

16:32

using AI as a kind of sounding board.

16:35

I think of it as almost the

16:37

infinite library. I can shout into it

16:40

and I'll get these echoes. I can say, help me with

16:42

these thoughts. What have I missed? What's going on? Have

16:44

I said it right? The

16:46

more formulaic, the easier it

16:48

is to outsource more of this task. It's

16:50

like having, as many people have said, infinite

16:53

interns at your disposal,

16:55

tireless. junior co -workers. But

16:57

crucially, you still

16:59

have to be in the driving seat when it comes

17:01

to the quality control, when it comes to the purpose,

17:03

when it comes to the intentions. And

17:06

potentially, it's quite dangerous to outsource

17:08

some of these decisions because you risk

17:10

not only kind of de -skidding yourself

17:12

and if you like diluting the

17:14

brand that is you, but you also

17:16

risk developing the very important kind

17:18

of human knowledge about what is

17:20

a better or worse email in

17:22

the first place. What is a

17:25

message that will differentiate you from

17:27

other people? What do you really

17:29

think and mean? And

17:31

so the challenge for me, which

17:33

I relish, is how organizationally, individually,

17:35

we can double down on these

17:37

things where it's essential we stay

17:39

in the loop while doing exactly

17:41

what you're saying and asking this level two,

17:43

this level three question, which loops do

17:45

I want to be in where and how?

17:48

And what should I gratefully outsource

17:50

in order to greater leverage

17:52

my own capabilities? I saw

17:54

this amazing piece of data that came out

17:56

last week. I think it was related to Canada,

17:58

but it may not have been that essentially 100 %

18:00

of professionals are looking

18:02

to switch jobs this coming

18:05

year, which is an astounding

18:07

number. And I'm throwing this

18:09

at you because I'm thinking about what

18:11

we just talked about. I'm thinking

18:13

about the fact that I've had very

18:15

recent conversations with people who have

18:17

very confidentially said to me something akin

18:20

to, do I even work

18:22

for a living? And what

18:24

they meant was that their days are

18:26

filled, their calendars are filled with meetings

18:28

and email and responses, but essentially the

18:30

self -reflection has led them to a

18:32

place of thinking that, I don't even

18:34

know if this is a job where

18:36

all I'm doing is almost meet in

18:39

the room, whether it's Zoom or physically, but

18:41

I'm just in meetings all day, we're sending

18:44

emails back and forth. I don't actually even know

18:46

what I accomplished in the past couple of

18:48

years, not in a way of which they're feeling

18:50

that they haven't done any work, they're doing

18:52

their job at performing exceptionally well. But

18:54

the work almost falls into what they would

18:56

call those bullshit jobs or fake jobs or

18:58

like, what do we need this for? The

19:00

other example that I constantly think about is

19:03

Elon Musk walks into Twitter at the time

19:05

and gets rid of all these employees and

19:07

everyone goes, how can you do that? And

19:09

sure, there have been problems. I think a

19:11

lot of them are more political and content

19:13

based rather than infrastructure based, but I mean,

19:15

it doesn't seem to go down. It doesn't

19:17

seem like they haven't added features. In fact,

19:20

it seems like they've added more features since

19:22

him than they could in a quarter or

19:24

two before. And all of

19:26

this just gets me thinking of how

19:28

we think about work. And as we

19:30

see this proliferation of AI, I think

19:32

you and I sit in the somewhat

19:34

privileged place where we're almost paid to

19:36

think about it and ask, but the

19:38

vast majority of people are running

19:40

their days and thinking, well, if my

19:42

meetings are just meetings and now those

19:44

meetings are being taken over by generative

19:46

AI, they're creating the transcripts, they're creating

19:48

the follow -ups, what are we actually doing?

19:51

Absolutely right. And the great anthropologist David

19:53

Graber coined that phrase, bullshit jobs, in

19:55

his book, an essay of that name. And

19:58

he was making a broader point

20:00

about the kind of information society that

20:02

it generates various forms of busy

20:04

work, but also that it generates a

20:06

kind of shadow world when on

20:08

the one hand your job is to

20:10

increase shareholder value is to build

20:13

relationships. But on the other hand, in

20:15

a way you do it through

20:17

the other kind of performative data moving

20:19

or shifting data around. It's

20:21

very interesting, of course, to look

20:23

at the jobs that will not

20:25

be replaced by AI anytime soon,

20:27

the ones that during the pandemic

20:29

kept us all alive, people delivering

20:31

food or growing food or cleaning

20:33

the streets or working in hospitals. Although

20:36

computers are making Jobs

20:38

in those areas are perhaps

20:41

more efficient or more challenging or

20:43

better or worse for that

20:45

matter. We are not seeing any

20:47

lesser -than -eat for nurses, for

20:49

physicians, for care workers, for

20:52

teachers, for headdressers, for

20:54

plumbers, for electricians and so on. Of

20:56

course, what these jobs have is an intimate

20:58

relationship with either the physical world or the

21:00

people in the physical world or both. And

21:03

so we do see in the kind

21:05

of knowledge sector, One of the strangest things

21:07

I feel about AI is that because

21:09

of the very nature of generative AI, it's

21:12

extraordinarily good at

21:14

doing creative tasks,

21:17

at creating poetry, at

21:19

creating videos, at creating images, and

21:21

it cannot yet write a great play or

21:23

write a really great book, but it can

21:25

write a pretty darn good article, it can

21:27

write a passable poem, it can create a

21:29

photographic image with something that looks as good

21:31

as the work of a world -class photographer. But

21:34

yet, These are jobs, these

21:37

are areas in which people do not

21:39

want AI to be doing it. They don't

21:41

want to buy the product of an

21:44

AI when it's not commodified. Of course, they

21:46

do if it's background music, but

21:48

people want to have a relationship with other

21:50

humans. And this then is a second key for

21:52

me. On the one hand,

21:54

there's still the jobs that people have

21:56

to do because lifting someone out

21:58

of a hospital bed or seeing to

22:00

their wounds is still a human

22:02

job, even if an AI is doing

22:04

the blood test. But secondly, there's

22:06

the jobs that people want to be

22:08

done by humans. They want their

22:10

children to be looked after by a human,

22:12

but they want their novel to have been written

22:14

by human. They want their movie to have

22:16

human actors in for it to be an act

22:18

of connection and cultural self -reflection. But if you're

22:20

not in one of those two categories, if

22:23

people don't care if an AI

22:25

is doing your job as long as

22:27

it's done well and they don't

22:29

intrinsically have any kind of physical or

22:31

bodily aspect to it, then yes,

22:33

that aspect of the job It looks

22:35

like something where neither of these

22:37

two reasons pertains. It's neither essential that

22:39

a human does it nor highly,

22:41

if you like, kind of ethically desirable

22:43

that a human does it. So

22:45

I do think we're going to see

22:47

and in some ways, maybe there

22:49

is something healthy about the organizational self

22:51

-examination that will have to take place.

22:54

When people and organizations are

22:56

saying, okay, where do we

22:58

need people in terms of

23:00

the morality, the ethics, the

23:02

relationships, the worth, the skills

23:04

that people have that machines

23:06

don't, where do we truly

23:08

need people? And we have people

23:10

themselves saying, well, look, if I'm doing

23:12

stuff that an AI can do

23:14

just as well, if I'm using it

23:16

to kind of cheat on my

23:18

job or show up half the time,

23:20

that's sending me a very strong

23:22

signal that my work is not perhaps

23:24

either intrinsically or extrinsically worthwhile. I

23:27

would just draw in the idea

23:29

of students because I deal a lot

23:31

with universities and there's obviously a

23:33

huge crisis around the fact that AIs,

23:35

that large language models, can generate

23:38

very, very good responses to many examined

23:40

questions. But of course,

23:42

I don't think this has people saying,

23:44

therefore, there's no point someone studying

23:46

for a degree. There's no point someone

23:48

learning. They're saying that in terms

23:50

of assessment and how we think about

23:53

learning, we have to really shift

23:55

our ideas. We have to find ways

23:57

to be testing people as thinkers, as

24:00

critical and creative thinkers, as prompters of

24:02

AI, as people who work with and through

24:04

these systems, as collaborators and so on. And

24:06

this can be kind of prefigures a

24:08

lot of what is going to happen,

24:11

I think, in workplaces where to be

24:13

optimistic for a second, a

24:15

lot of people will be

24:17

challenged to really think hard about

24:19

where, how, and why people

24:21

contain various forms of tacit knowledge.

24:24

people perform various tasks

24:26

that AIs can't. And

24:28

of course, AIs highly commodified, right? So

24:31

ultimately, if

24:33

you're just using AI to

24:35

run your company, you're probably

24:37

going to be eaten up and

24:39

spat out by someone else somewhere

24:41

else who's doing it cheaper and

24:44

faster at scale. But there is

24:46

some conversation and rhetoric around that

24:48

type of, I think it's almost

24:50

the binary way of stating it.

24:52

What I would mean is, If

24:54

you think about, let's say, AI

24:56

chatbots for nursing. So

24:58

suddenly, if you have privilege, you'll have

25:00

a nurse that can take care of you

25:02

or caregiver, and they would be using

25:04

AI or generative AI as some type of

25:06

supplement to make results better, more effective.

25:08

What have you? But there's also a

25:10

larger swath of people who are thinking about

25:12

this and very worried that suddenly what we'll say

25:14

as well, you don't have healthcare insurance or

25:16

you can't afford it or you're not important. So

25:18

you just get the chat bot. You get

25:20

the chat bot and say to it, my blood

25:22

pressure is high. I didn't take my medication.

25:24

I don't feel well. It interacts with you and

25:26

then it will level it up in terms

25:29

of needs. So we look at that and go,

25:31

well, isn't that phenomenal? I

25:33

think some of the thinkers are saying,

25:35

no, that's actually gonna create a greater

25:37

divide. and that the challenge isn't

25:39

that you have it or you don't.

25:41

It's that we're using it to replace or

25:43

hold others back in a world where

25:45

those who can afford more can have more

25:47

access. Yes, and we are

25:49

seeing a version of this already. I

25:51

deal a lot with kids. And

25:53

anecdotally, I will certainly find that if

25:55

I'm dealing with the children of

25:57

senior people and tech companies, their

26:00

education involves a lot of

26:02

exercise, yoga, meditation,

26:05

good quality food. beautiful environments and so

26:07

on, they are not spending 24

26:09

hours a day on devices partly because

26:11

they're being educated not to. And

26:13

at the other end of the spectrum,

26:16

people who have less privilege are

26:18

far more likely to be, as it

26:20

were, in an area where their

26:22

time is being turned into other people's

26:24

money through algorithmic means. They're watching

26:26

adverts, they're less able to afford

26:28

fresh nutritious food, they're making longer commutes.

26:30

And we do see a version of

26:33

this. And there is a very powerful,

26:35

but I think very double -edged argument

26:37

that says it's the better than nothing

26:39

argument. There's not enough money and

26:41

resource to go around. So it's better that

26:43

you chat to a psychotherapist bot that

26:45

gives you something than that you join the

26:47

three -year wait to talk to a human

26:49

psychotherapist. Or that we make the effort

26:51

that we're not going to do that because

26:54

we know the result. Or that

26:56

we're not going to just give

26:58

everybody in a low -income public school

27:00

system a AI bot math tutor. And

27:03

of course, we are creating

27:05

this strange world. And

27:07

it's very double -edged because on the one hand,

27:09

when I go and talk to students from lots

27:11

of different backgrounds, they say, I'm

27:13

struggling with my course and

27:15

I find AI really useful. It

27:17

explains stuff to me, it

27:19

listens to me, the

27:21

instructors try their best, but they don't have

27:23

the time. I'm dealing with overwhelming information. This

27:26

tool is amazing because it helps me

27:28

take these this overwhelming information, and it

27:30

helps me understand it. And yes, it's

27:32

pretty tempting. It could write an

27:34

answer for me, but I try quite hard not

27:36

to have the answer written for me. But on the

27:38

other hand, of course, that means you can have

27:40

a new model where you can effectively say, well,

27:43

this is good enough, right? And

27:45

I have no answer to this because

27:47

this is partly about what, as

27:49

a society, people value, people think they

27:51

owe to one another, about where

27:53

and when and how we think we

27:55

do or don't owe people our

27:57

time, our attention, our value, and

28:00

it's deeply also about

28:02

human levels of comfort

28:04

and discomfort. Now,

28:06

we were already seeing whole

28:08

swathes of the workforce in, for

28:11

example, jobs where people are

28:13

task -grabbits or in warehouses or delivery or

28:15

whatever and delivery where you are managed

28:17

by algorithm, where in effect, to be

28:19

an Uber driver, and plenty of people

28:21

are very grateful for the ability to

28:23

earn money in this way, but... They

28:25

are managed by algorithm by and large.

28:27

They log into the app. They are

28:29

casualized. There's very few humans there. Structurally

28:32

speaking, this has all kinds

28:34

of model hazards, people's dignity, people's

28:36

health, people's right of appeal,

28:38

and so on. Now, in

28:40

its own way, this looks

28:42

a little bit like other

28:45

eras in which we have

28:47

seen vast new kinds of

28:49

automation and mass production coming

28:51

in when we've had the

28:53

ability to, as it were, make

28:56

automated large amounts of things

28:58

that used to be bespoke.

29:00

And that can both remove

29:02

certain inequalities and create enormous

29:04

potentials for exploitation. And

29:07

we've had to have legislative interventions

29:09

around those. We don't have children working

29:11

in factories in a lot of

29:13

countries we do in some. We provide

29:15

unions to those who might be

29:17

uneducated or might be a larger workforce.

29:19

We have universal education and universal

29:21

healthcare in some countries, although not in

29:23

others. These have been defined as

29:25

rights. So there's going to

29:27

be a real battleground here. Do people

29:29

have a right as is being

29:31

enshrined in legislation like the EU's AI

29:34

Act, which for all its flaws

29:36

is a very bold attempt to talk

29:38

about rights? Do people have a

29:40

right to have things explained to them?

29:42

to have explicability? Do people have

29:44

a right for an algorithm not to

29:46

monitor certain aspects of their kind

29:49

of biometrics because that's too intrusive? Do

29:51

people have a right to healthcare,

29:53

regardless of their genetic profile, when someone

29:55

might be too genetically risky to

29:57

ensure? And again, we've seen this

29:59

for the insurance legislation that insurance decades,

30:01

by the way. Of course, insurance companies

30:03

are simply not allowed to, as it

30:06

were, kind of micro -profile you and

30:08

deny you these things. So these debates

30:10

have been had before, but... But we're

30:12

doing it on the fly. We're rebuilding

30:14

the ship while we're sailing. And

30:16

the trouble with rebuilding your ship while

30:18

you're sailing is you often only tend to,

30:20

so to speak, pass the relevant legislation

30:22

when you watch a large chunk of your

30:24

ship drop off and a whole bunch of

30:27

people in the water screaming at you,

30:29

I'm mixing my metaphors now. So

30:31

I have no answers to this.

30:33

What I do tend to advocate

30:35

just in general is that These

30:37

ethical concerns and these very baseline

30:39

concerns about what are our beliefs about

30:42

people, about rights and so on,

30:44

are present in discussions, are present in

30:46

the boardroom, and I don't think

30:48

it's just idealistic because I think this

30:50

huge kind of reputational long -term hazard

30:52

to organisations for being on the

30:54

wrong side of these things. But when

30:57

we make decisions about data, we

30:59

make decisions about people's lives, people's people's

31:01

childhoods, people's rights, and we have

31:03

to be very careful indeed Just at

31:05

the idea of we can hand

31:07

it over to systems and it's all

31:09

good. Equally we have

31:12

to be pragmatic and so sometimes

31:14

the solution there is one to

31:16

try and listen to the end

31:18

users try and talk to people

31:20

to try and talk to people

31:22

rather than to talk for them.

31:24

And spend as much time as

31:26

possible with people talking about their

31:28

experiences of healthcare education work and

31:30

management and see what you can

31:32

do about. actually meeting them

31:34

where they wish to be met. What

31:36

a volley of thought over to you

31:38

and preface it by saying I may

31:40

not formulate the words as effectively as

31:42

I want, but it's something that came

31:44

out in the work I'm doing in

31:46

reading wise animals. It was a note

31:48

that I kept the feeling was reoccurring

31:50

to me, which is this idea that

31:53

maybe what we're seeing is a devaluation

31:55

of knowledge workers, meaning our perception of

31:57

what knowledge work is when you put

31:59

general AI into me, isn't about AGI

32:01

or super intelligence. It's about the fact

32:03

that it's pretty damn good. It can

32:05

actually replicate a lot of the repetitive

32:07

tasks that executives or knowledge workers do,

32:09

and it does it in a great

32:11

way. And that perhaps if

32:13

we recognize that and see that, that's

32:16

the cataclysmic problem. Like, oof, got

32:18

this advanced degree. I find myself in

32:20

some level of what I would

32:22

call superiority. I manage people, tell them

32:24

what to do, get ridiculous sums

32:26

of money for it based off of

32:28

them doing the actual hard labor.

32:30

There's all of that. And

32:32

suddenly, it's the wizard. It's the curtain being

32:34

open that says, actually, the work that you're

32:37

doing is pretty easy to replicate and to

32:39

do. I think there's

32:41

a lot of truth in that.

32:43

I think that ties in

32:45

with a lot of social resonances,

32:47

if you like, that are

32:49

still in process. I

32:51

feel that there's a connection

32:53

there between this intuition and

32:55

A lot of the tone

32:57

of populism, when people have

32:59

been saying a version of,

33:02

we've had enough of experts, of

33:04

these finally educated people

33:06

don't know what's really going

33:09

on. The technocrats,

33:11

these people who are so clever with

33:13

their degrees, they're not delivering. They

33:16

make promises, they claim they're so clever,

33:18

but they're not. I'm not

33:20

pronouncing on the rightness or wrongness of

33:22

that, but it's a vibe. And

33:24

certainly, once you can

33:26

go to chat GPT or

33:28

Gemini or whatever and replicate the

33:30

work that five years ago

33:32

you were being told was so

33:34

fancy and unique that you

33:37

needed seven years of education and

33:39

a lot of letters after

33:41

your lane to do it. It

33:43

does something to the scarcity

33:45

value of that work. And

33:47

of course, a lot of members of

33:49

the academy of the elite have got a

33:51

lot of, if you like, moats and

33:53

draw bridges around the place. It's highly specialized

33:55

for categories. Lots of

33:57

qualifications that are kind

33:59

of keep out signs.

34:03

I am already

34:05

suspicious, I think, of people

34:07

who deal in jargon, of people who deal

34:09

in abstraction. Sometimes you need to, of

34:11

course, especially if you're operating in rocket science

34:13

or biochemistry, of course you need to

34:15

deal in jargon. But when it comes to

34:17

explaining why these things matter and why

34:19

things are important, I think

34:21

a lot of what passes for intellectualism.

34:23

There is a little bit of Wizard

34:25

of Oz going on. We

34:27

are seeing, I think, a much greater

34:30

premium going on to the soft skills

34:32

of empathy, of collaboration, or if you

34:34

like, their kind of mirrored sides, their

34:36

dark aspects, charisma, whipping

34:38

people up into a frenzy, manipulation,

34:40

charisma, and empathy are not

34:42

necessarily good things. They are enabling

34:44

virtues rather than cardinal virtues. So

34:47

I think you're absolutely right about

34:49

this. And I think One

34:51

of the interesting things for me is

34:53

where we flip over from the things

34:55

that people are suspicious of, like a

34:58

bunch of experts talking in a kind

35:00

of high flute and language about things

35:02

that may or may not actually benefit

35:04

others, towards the stuff perhaps that they

35:06

do love, beloved stories, beloved

35:08

performers, singers, songwriters, Taylor Swift,

35:10

who is in one sense,

35:12

absolutely a member of the

35:14

hyper elite, an incredibly talented

35:16

performer and songwriter, incredibly charismatic,

35:18

global figure. But at the

35:20

same time, someone who

35:22

speaks in a common idiom and generates

35:25

great popular works of art, our people

35:27

are not too happy, quite rightly, about

35:29

her image, her songs, her lyrics and

35:31

her ideas being ripped off by AI. And

35:34

people like that are leaders in

35:36

the movement against these synthetics. Someone like

35:38

Nick Cave, who is perhaps a

35:40

more kind of elitist, but nevertheless has

35:42

a huge audience. I love

35:44

video games as an art form. I've

35:46

written a book about video games. I

35:48

love them partly because the great video

35:50

games are demonstrably incredible works of art

35:52

and also wildly popular and accessible. And

35:55

so I do think that

35:57

suddenly a lot of people who

35:59

maybe had cushy lives but

36:01

whose status was unproblematic to them

36:03

now have to sweat quite

36:05

a bit and find new ways

36:07

to sing for their suppers. And

36:10

this is a really interesting

36:12

moment and a dangerous moment

36:14

perhaps because very easily a

36:16

meaningful critique can flip over.

36:19

into kind of demagoguery, rabble

36:21

rousing, and the devaluing of

36:23

the very precious thing that

36:25

lies behind the jargon, which

36:27

is knowledge that has been

36:29

tested and hard won, and

36:31

that is coupled to a

36:33

really sincere effort to understand

36:36

the world better, to understand

36:38

people better, to make it

36:40

better. I have a quick

36:42

story that I think illustrates this, which I'd

36:44

love just you to react to, which is

36:46

I was giving a keynote presentation. And

36:48

I was told for the morning keynote

36:50

that I need to come in a bit

36:52

later than I normally would to make

36:54

sure everything works because they were doing their

36:57

in -camera board meeting for this association of

36:59

a very regional but large industry. And,

37:01

you know, I'm waiting in the hallway and someone

37:03

says, oh, it's okay. Like, clearly you're not. Somebody's

37:05

going to spy or write articles about this. Go

37:07

in and do your little setup. So

37:10

they were talking primarily about the

37:12

five major issues facing this large industry

37:14

within this one specific geography. And

37:16

they're talking and I'm doing my thing

37:18

and somewhat laughing a little bit

37:20

because one of the things I will

37:22

do to just demonstrate how generative

37:24

AI makes us feel like it's intimate

37:26

is the way it types out

37:28

and thinks and does all these things

37:31

that are very fake, but they're

37:33

sending signals to us that it's just

37:35

for us or intimates, which I

37:37

think is another conversation point. And

37:39

what I'll typically say is you

37:41

are an expert in industry XYZ.

37:43

You are specialized in region. What

37:46

are the five major issues facing

37:48

that industry? And I do

37:50

this in what I call virgin mode, meaning it's not

37:52

using my custom interactions. It's not using the legacy. It's

37:54

just pure log and see what it does. And

37:56

it was almost exact to the

37:58

issues that they were discussing in camera

38:00

as if they had this deep

38:02

and it was using acronyms specific to

38:04

laws that are happening and that

38:06

are being enforced in their region. And

38:08

I just remember the jaws being

38:10

completely dropped because it wasn't even your

38:12

knowledge. It was the understanding of

38:14

the area, which again, you would think

38:16

is I have to pay this

38:18

association fee. I'm paying them to advocate

38:20

on my behalf and yet AI

38:22

in its own little simple way was

38:24

able to pull out every single

38:26

issue. It reminds me of your email

38:29

anecdote. You don't have to

38:31

do the first draft. The first draft is

38:33

there waiting for you. And so the skill

38:35

becomes what can you do with that first

38:37

draft that another AI or a person on

38:39

the street or even a royal organization can't

38:41

do. And of course one thing people know

38:43

about is their own organization in real depth

38:45

what it's capable it is our five things

38:47

you could do but you know that one

38:50

of those. Is gonna work and

38:52

that for those these people you've got

38:54

content or you need these people but

38:56

it is it's that first drafts thing

38:58

and I think. It

39:00

makes a lot of people very uncomfortable and it

39:02

should. Because a

39:04

subsidiary question is, well, is there

39:06

a benefit nevertheless to that group

39:08

of people coming up with these

39:10

five things, even if an AI

39:12

could have come up with it

39:14

for them? Is that process intrinsically

39:16

valuable? And I would

39:18

suggest that sometimes there is intrinsic

39:21

value to that process, but you

39:23

can absolutely have a kind of

39:25

AI pre -read that does some of

39:27

the heavy lifting for you. And

39:29

interestingly, some of the work I

39:32

do is often trying to come

39:34

up with, if you like, more

39:36

interesting structures for board meetings, for

39:38

debates, in order to get the

39:41

most out of people and systems.

39:43

And I put a big emphasis

39:45

on pre -reading, on people contributing

39:47

in advance of that, inputs from

39:49

quite diverse perspectives to big picture

39:51

prompts. That's quite the Jeff Bezos

39:54

pre -meeting strategy. Absolutely. And I

39:56

think Amazon's strategy on their meeting setups

39:58

is cognitively very literate, if you sort

40:00

of mean. because it's putting a huge

40:02

emphasis on making the most of your

40:04

people and not getting the people to

40:06

behave in a machine -like way. And

40:08

again, another famous thing that Amazon does

40:10

is you tend to get your data

40:12

from APIs yourself rather than from a

40:14

slide deck produced by somebody else. Because

40:17

then you are an interactor and a

40:19

questioner rather than just the passive recipient

40:21

of someone else's cam presentation. Talking

40:23

to law firms, they worry a lot.

40:25

Sorry, I'm getting carried away here. No,

40:27

please. About the fact that On the

40:29

one hand, it is potentially transformative for

40:31

them. That AIs can just do

40:33

a lot of legal grunt work, a lot

40:35

of paralegal or junior partner type work, even very

40:37

fast and very efficiently, and with a moderate

40:40

to high degree of accuracy. On

40:42

the other hand, how are you going

40:44

to have the senior partners of

40:46

tomorrow if they haven't done the time

40:48

working down in the nitty gritty

40:50

of contracts and precedents and draftings and

40:52

really honing their skills? I

40:54

don't have an answer to that, but

40:56

one thing I always point out to them

40:59

as well is it's not only that.

41:01

It's also that because everybody has this technology

41:03

and the man or woman on the

41:05

street has this technology, suddenly the legal environment

41:07

is going to become incredibly noisy because

41:09

I can get a bot to file a

41:11

suit for me. I can get a

41:13

bot to generate precedence and another bot might

41:16

read it and suddenly we're in this

41:18

environment where really good advice and really good

41:20

strategy is also going to be about

41:22

dealing with the level of what David Foster

41:24

Wallace called in the context of television

41:26

total noise. just this

41:28

ultra -noisy, algorithmically enhanced environment

41:31

where your skills to cut

41:33

through that strategically and

41:35

intellectually becomes sort of hyper

41:37

-impled. I had two thoughts

41:39

on this as you were talking, and

41:41

it's thoughts that I've put forward,

41:43

but I think they're just germane to

41:45

the conversation, interesting stuff to discuss,

41:47

which is the answer would seem to

41:49

be almost twofold. One

41:51

is the word that you haven't used

41:53

that I used often is commune.

41:55

The value is in the commute. The

41:57

value is in the fact that

42:00

we are in the room together and

42:02

we're using whatever technological tools we

42:04

have. But clearly, if we learned anything

42:06

from COVID and the emergence of

42:08

AI, it's that, well, we said shopping

42:10

malls were dead long before COVID.

42:12

Suddenly, everyone's going shopping malls. We recognized

42:14

in the removal of the physicality

42:16

how much we require it. We saw

42:18

people being arrested for gathering in

42:20

parks and things like that. So clearly,

42:23

and we're seeing this just in the pushback of

42:25

being a keynote speaker of events, of people going

42:27

to physical movie theaters. I don't think that's going

42:29

to last forever, but they're still doing it now. And

42:32

then the other thought is maybe

42:34

it's more of an insane provocation

42:36

that I've made is what we

42:38

might need to do is make

42:41

these advanced degree professions trades. I

42:43

mean, what if we approach the

42:45

education of a doctor, lawyer, engineer, and

42:47

others as a trade? Because we

42:50

know you're going to be supported in

42:52

healthcare by a healthcare practitioner, a

42:54

nurse practitioner. We know we're going to

42:56

be using AIs. The brunt

42:58

of work and knowledge that typical doctors

43:00

have done historically is no longer there.

43:02

And what if we did that? Perhaps

43:04

we would have an influx of more

43:07

people who actually wanted to care for

43:09

other people. Perhaps we would have a

43:11

profession of legal where it isn't just

43:13

my wordsmith, outward smithing, your wordsmith in

43:15

a court of law, but actual thinking

43:17

about what it means to protect both

43:19

the business and the interest. Those

43:22

were my two paths and they're very divergent.

43:24

So feel free to tack on where you want.

43:26

Well, I'd be interested to trample them together

43:28

because I do think the people in the room

43:30

or the connection between the people because, of

43:32

course, for certain kinds of people, a

43:34

combination of the kind of online,

43:36

the technological and the in -person can

43:38

build relationships, build deeper relationships and build

43:41

relationships with people who can't always

43:43

physically get to the room. I

43:46

particularly think in the context of

43:48

things like medicine and healthcare, which

43:50

is an area I know well because of

43:53

family and friends who work in these areas

43:55

and work I've done, was very

43:57

much saying a move to war

43:59

is kind of problem -based. education, where

44:01

you're hands -on, where you're dealing with

44:03

practical scenarios because, among other things, it's

44:06

your ability to work in an interdisciplinary

44:08

team in a healthcare system in the

44:10

real world that has a profound effect

44:12

on patient outcomes. It's

44:14

about what happens in the

44:16

operating theatre, in the

44:18

consultation, or on the complex

44:21

care pathways and the

44:23

interactions between primary care physician,

44:25

secondary care physician, the

44:27

community support worker. the pharmacist,

44:29

the cardiologist, the hematologist, and

44:32

so on. What this

44:34

suggests to me is that it's

44:36

absolutely right that the interpersonal

44:38

and communicative skills and the ones

44:41

that are to do with

44:43

people's ability to understand each other

44:45

and communicate well, very

44:47

inspired by Robin Dunbar's another work.

44:49

He wrote a book recently

44:51

with Tracey Camilleri and another bunch

44:53

of business experts. talking about

44:55

the implications of his sociological work,

44:58

the famous Dunbar number of 150

45:00

people that we can perhaps

45:02

form kind of emotionally meaningful relationships

45:04

with. What this implies for

45:06

organisations, it puts an

45:08

enormous emphasis upon empathy

45:10

building, upon communication, upon morale,

45:12

upon these spaces that

45:15

permit constructive disagreement and so

45:17

on. And I would say

45:19

that what we have in a trade, of

45:21

course, is a model of

45:23

education where if it's done well,

45:25

you are having people learning on

45:27

the job in a cohort who

45:30

are mentoring each other, learning from

45:32

experience other people, solving practical problems

45:34

with their hands in the real

45:36

workplace. You don't learn to be

45:38

a carpenter in theory. You don't

45:40

learn to be a great plasterer in

45:42

theory. You learn on the

45:44

job. And I think organizations

45:46

and if you like elite workplaces

45:49

should look a lot more like

45:51

this. because what does it mean

45:53

to have a bunch of different

45:55

people with different skills and experiences?

45:57

A, build real rapport and really

45:59

approach a problem and grasp that

46:01

problem and grasp its complexities and

46:03

grasp the experiences and not just

46:05

its algorithmic representation. And then

46:07

what does it mean for them to get down to it,

46:09

communicate well, communicate frankly, adapt,

46:11

change, maintain high morale,

46:13

deliver value to people in the real world,

46:16

deliver the stuff that people value,

46:18

which is going to be

46:20

increasingly about what algorithmic systems can't

46:22

or compensating for the weaknesses

46:24

of algorithmic systems, which will tend

46:26

to be very poor at

46:29

reframing, critical thinking, which will tend

46:31

to be very poor at

46:33

truly understanding that which cannot be

46:35

quantified. And of course, at

46:37

delivering a meaningful interpersonal relationship. So

46:39

there's a lot be learned from that.

46:41

And maybe there's a great case to be

46:43

made in business skills and elsewhere for

46:45

working more closely with industry, for working more

46:48

closely in graduate work. in part time

46:50

work, in not just a bunch

46:52

of people at the age of 17, 18,

46:54

19, 20 going and having a largely

46:56

abstract elite education. And then it

46:58

being assumed that they can go and

47:00

work in knowledge jobs and be paid

47:02

very well for that. And

47:04

of course, you pointed out

47:06

earlier that there is quite a

47:08

lot of damage being done

47:10

to people's morale, people's sense of

47:12

self by this feeling that

47:14

they are cut off from meaning

47:17

in their work, that they

47:19

are not. in a team of

47:21

people communicating meaningfully with each

47:23

other and with customers, clients, users,

47:25

society that they're in a kind

47:27

of abstract information bubble where they're busy

47:30

training the systems that will replace

47:32

them. That is not good for organisations,

47:34

not good for people, not good

47:36

for society. And we are seeing

47:38

also I think huge growth in higher

47:40

education in precisely what you might call

47:42

the informal areas. Universities that

47:44

are more affordable, that are more flexible. that

47:46

will let you skill up while working in

47:48

the real world while being paid or having

47:50

other skills that will come in when

47:52

you're older. We're saying more professional education, more

47:55

universities that say to people, yeah, you're in

47:57

your 30s, you're in your 40s, you've got

47:59

kids, you've got dependents. We'll train you

48:01

up to be a nurse. We'll train you

48:03

up to be an engineer. We'll train you

48:05

up to be an electrician. We'll train you

48:07

up while you do other things. And

48:09

this learning will not be the

48:12

kind of learning that an 18

48:14

-year -old who spends their entire time

48:16

Interrupting with the screen might expect.

48:18

Let's talk a little bit philosophically

48:20

about what you're seeing in terms

48:22

of artificial intelligence. Do

48:24

you see the emergence of

48:26

AI vastly different from other

48:28

technologies? And the reason

48:30

I'm submitting this idea to

48:33

you is because I've been

48:35

involved in technology since personal

48:37

computers came in, early

48:39

80s. My sentiment having gone

48:41

through the cycle and developed successful businesses

48:43

and others in every cycle of

48:45

this is I don't even think we're

48:47

thinking about this 1 % correctly my general

48:49

sentiment and yet others are kind

48:51

of passing it off like it's just

48:53

the technology and we adapt as humans

48:56

and it's okay and I have a

48:58

real troublesome time with that, and I

49:00

self -admit that I have an even

49:02

harder time having the debate and podcast

49:04

with them because they're typically business executives

49:06

who aren't really thinking about it.

49:08

They're like, oh, well, just adapt. It's

49:10

technology. And I have this

49:13

blood out of my brain look of

49:15

complete fear where I think as we

49:17

develop this, and by the way, whether

49:19

or not we hit AGI or superintelligence,

49:21

I'd even make the argument that if

49:23

it stopped right now, it's deadly dangerous

49:25

in terms of what it can do

49:27

already. I'm inclined to agree with you

49:29

really. I try to resist hype, not

49:32

because I think things should be minimized,

49:34

but because I think hype often points in

49:36

the wrong direction with too simple a

49:38

story. It tells, as

49:40

people like Brown Merchant have pointed out in

49:42

their work, hype is a kind of marketing

49:44

exercise even when it's negative. This

49:46

is so powerful and so big, you can't

49:48

afford to miss out on it. It's so vast,

49:51

it's so important, it's so epoch changing, you've

49:53

got to just buy in and go with it.

49:55

That sort of diminishes our agency. But

49:57

I think it's huge and I am

49:59

bewildered by people who minimize in the

50:01

sense of saying, ah, it's, you know,

50:03

same old, same old. You know, I

50:05

have a system in my pocket that

50:07

I can ask any question in the

50:09

world in natural language and get a

50:11

fairly sensible answer. Perhaps even more worryingly,

50:13

sometimes that answer will be misleading or

50:15

utterly wrong and yet it will be

50:17

incredibly fluent and confident. You can draw

50:19

me a picture, it can make me

50:21

a video, it can imitate a person,

50:23

it can copy a book and so

50:25

on. and so on. And

50:28

so things that were a few years

50:30

ago, so far as most people are

50:32

concerned, uniquely the province of humans or

50:34

highly skilled humans, are now potentially commodified

50:36

and out there. We've suddenly got a

50:38

system that can write you a good

50:41

essay in almost any humanities subject, such

50:43

that you can't really tell whether it's

50:45

an AI or a pretty well -informed human.

50:47

That is crazy. We've got a system that you can pop

50:49

a book into it and say, what's this book about? Pretend

50:52

to be the author and it will do that. I

50:54

mean, this is This is so

50:56

big, it's hard to know what to do with it.

50:59

I don't think we should do what a lot of AI

51:01

people say we should do with it in just sort of,

51:03

I don't know, submit to the utopian. But

51:05

equally, it is a

51:07

massive deal. And of course, it

51:09

feeds upon previous steps in

51:11

the way technology does. So

51:13

it's predicated upon vast amounts of data.

51:16

So we have turned our

51:18

world, or we have generated, as

51:21

it were, a machine readable version

51:23

of our culture. and all aspects of

51:25

our world. And now machines are

51:27

reading it and analyzing it. An enormous

51:29

power comes with that. The good

51:31

thrill. I don't even know. We

51:34

generated the ability to. And I've been

51:36

in San Francisco recently. I was toddled

51:38

around the place and self -driving a

51:40

vehicle. Watch them a little bit like

51:42

I'm watching a new species in the

51:44

wild. Watching their behaviors, watching them flock

51:46

and swarm and have this kind of

51:48

It looks more like an ecology. They're

51:50

doing their little things on the road and

51:52

they don't interact with the road system like

51:54

normal human driven vehicles. But

51:56

they're out there. I know there's remote people

51:58

monitoring them. It's

52:01

really big and what we want

52:03

to do, I think, is be

52:05

thinking about it and trying stuff,

52:07

keeping our wits about us and

52:09

experimenting and celebrating and warning and

52:11

getting specific and not just lurching

52:13

towards a simplified story that says,

52:15

oh, no big deal or such

52:17

a big deal, all our problems

52:19

have dissolved. And this, of

52:22

course, is what people find difficult psychologically. We

52:24

tend to express our tribe.

52:26

We tend to cope with the

52:28

world by expressing allegiance to

52:30

an oversimplified narrative. It's all good.

52:32

It's all bad. New technology is great. It's awful. It's

52:35

going to save us. It's going to doom us. All the

52:37

companies are bad and cynical and awful and all they

52:39

want is profit. It's the worst thing ever. And

52:41

this doesn't make sense to me. I

52:43

want to be there with people just

52:45

talking about what they're doing, what

52:47

they're learning, and how they're

52:49

learning and trying to be

52:52

useful, trying to give people thinking

52:54

tools for conducting meaningful experiments.

52:56

So you make a point of

52:58

the book -wise animals that what

53:00

we're seeing is how we

53:02

co -evolve with technology as individuals. And

53:05

we have seen this in terms of

53:07

how we've co -evolved with other species

53:09

over thousands of years. And I might

53:11

be very, very basic or pedantic in

53:13

this. I apologize if that's... The thought

53:15

that I was wondering is... So we

53:17

know that we've risen to the top

53:19

of the food chain, whether we should

53:21

or shouldn't is another philosophical Darwinism type

53:23

of conversation. But if we are introducing

53:25

this thing that you and I are

53:27

talking about now at that level, are

53:30

we still able to think about it

53:32

in this way? Are we still able

53:34

to think about it as a coexistence

53:36

of in and of its own self?

53:38

It becomes top of the food chain

53:40

in terms of just even how we

53:42

would define knowledge. So I've seen people

53:44

talk even about Some of the things

53:46

that have happened recently with Google and

53:48

their leap in terms of quantum computing,

53:50

that if you really break it down

53:52

what it's doing from a knowledge understanding, we

53:55

wouldn't even understand it. We couldn't

53:57

even talk, use words that we have

53:59

because of how it could operate

54:01

or think about the world and understand

54:03

it. What happens when we're not

54:05

co -evolving but suddenly we're not top

54:07

of the food chain either? So

54:10

for me, just my perspective, I

54:13

think that the most powerful

54:15

thing about understanding us as a

54:17

part of the planetary systems

54:19

that we evolved alongside. And

54:21

about technology as something distinct,

54:23

unique to our evolutionary heritage

54:25

that we evolved with and

54:27

think through, is that we

54:30

asked it in the same situation.

54:32

But the situation has always been

54:34

one in which as thinking beings,

54:36

we live alongside and

54:38

entwined with the incomprehensible,

54:40

the cosmic, the vast, There's

54:43

a nice line from a

54:45

philosopher whose name will have

54:47

to come back to me,

54:50

because we're doing this live and I

54:52

can't… we'll turn GBT it, exactly.

54:54

No, I refuse to do that for

54:56

now, but dig in myself. Go

54:58

ahead. Who suggests that partly what we're

55:00

doing in the context of Big

55:02

Data and AI is it's allowing us

55:04

to have almost a kind of

55:06

a direct interface within comprehensible complexities with

55:08

the kind of stochastic nature of

55:10

the universal. And certainly Google's Willow is

55:12

interesting. If you follow what's

55:14

been said by David Deutch and

55:16

others, there's a sort of non

55:18

-trivial prospect that quantum computation is

55:20

effectively indirect empirical evidence for the

55:22

existence of a multiverse, that

55:25

a calculation is effectively performed simultaneously

55:27

across all the superimposed versions of

55:29

reality. And then the waveform collapses

55:31

in this universe and we get

55:33

an answer. Now, my goodness,

55:35

that's an astonishing thing. But also, it's

55:38

like touching upon something

55:40

of such incomprehensible vastness that it shrivels

55:42

our minds to the size of

55:44

a speck. But that's what ought to

55:46

happen every time you look at

55:48

a plant or the night sky. We

55:51

are a tiny, tiny

55:53

part of a vast, largely

55:55

incomprehensible universe, and yet

55:57

we comprehend, and yet

55:59

we build tools that allow

56:01

us to look at its

56:03

grandest structures, megastructures in a

56:05

kind of gigaparsecs across. we'll

56:08

look down to the plank link. And

56:10

we'd already done this stuff before AI came

56:12

along. So AI is part

56:14

of us and our journey, I think. And

56:17

I think that the model for it

56:19

is not that it is something separate

56:21

from us, it is utterly dependent upon

56:23

us, utterly and wholly dependent upon us.

56:25

But we, in turn, are utterly and

56:27

wholly dependent upon this planet that we

56:29

live on all of our life. We

56:31

are dependent rational animals, to quote Alistair

56:34

McIntyre, a great philosopher of virtue ethics.

56:36

So our dependency has always been there. And

56:39

we've always been dependent. We may think we're

56:41

masters, we're not. That's a delusion. That's absolute

56:43

nonsense. Of course, we're not independent. What would

56:45

that even mean? But

56:47

we weave these vast collective

56:49

nets of power and

56:52

comprehension. They don't diminish our

56:54

ultimate dependency, but they

56:56

vastly increase our agency. And

56:58

yes, we have exponentially

57:01

increased our collective agency and

57:03

knowledge as a species, I think,

57:05

through the ability to interpret

57:07

and extrapolate and mobilize data. And

57:09

I don't even know what it

57:12

will mean. But I want for

57:14

myself to insist on it being

57:16

a revelation of something that has

57:18

always been true, that we lived

57:20

as part of this vast kind

57:22

of cosmic complexity and order, that

57:24

we are a tiny part of

57:27

it, and the miracle of us

57:29

in technology. is that somehow, somehow,

57:32

we managed to focus and

57:34

harness and render actionable

57:36

this immensity. You know,

57:38

the cliff notes on that is we are

57:40

going to see what was previously unseeable. Yeah,

57:43

all develop something that can see it

57:45

on our behalf. But remember, you don't

57:47

think about to hook in the microscope.

57:50

We've been sketching drawings

57:52

of the previously inconceivable.

57:55

or even the intuitions

57:57

that lead people towards

57:59

belief and a deity. Again,

58:02

intimations of the vastness, the

58:05

unknowable, the strangeness. That's what we do

58:07

as a species. We reach beyond, we

58:09

self -reinvent, we exceed, we are never content. It's

58:12

our curse and our blessing. You

58:14

talk in the book about the

58:16

need to have compassion, curiosity, and

58:18

humility. And I wanted to

58:20

give you an addition to that. That was

58:22

a gift, I think, to me just

58:25

in being in a moment, which is I

58:27

was at a conference here in Montreal

58:29

where I live. And very fortuitously, I live

58:31

in a city that also has Joshua

58:33

Bengio here, who is one of the early

58:35

developers of large language model LLMs and

58:37

just AI. He recently won the touring test

58:39

and he was being interviewed on stage.

58:41

And what he said that was so fascinating

58:44

to me, it's a thought I always

58:46

bring to these intellectual AI conversations is they

58:48

asked him what he would hope would

58:50

be built into these systems. And he said,

58:52

only one thing, self doubt. And

58:55

I thought it's such a beautiful addition

58:57

to what you had written in Wise

58:59

Animals, this idea that if these engines

59:01

are in some way super intelligent or

59:03

AGI or even in its current state,

59:05

if it just had some level of

59:08

self -doubt, that might provide a lot

59:10

of pathways for humans to interact with

59:12

it more effectively. Exactly right. And people

59:14

like Stuart Russell, absolutely exactly the same

59:16

thing. And I think they're quite right. And

59:19

doubt is a gift, of course, because

59:21

to doubt is to identify,

59:23

is to be interested

59:25

in that which we do not yet know,

59:27

is to admit we don't know everything

59:29

that our understanding is uncertain and yet to

59:31

try to move in the direction of

59:33

becoming less deceived. And it

59:35

strikes me as a very powerful thing

59:38

to work into AI. Now, there are

59:40

some people who say that the current

59:42

architecture of NLMs is, if you like,

59:44

inherently hallucinatory, inherently stochastic, and that says

59:46

maybe. But even then, how can

59:48

we make that a feature rather than a

59:50

bug? This incredible kind of creativity, this incredible

59:52

ability to match the extrapolation. I think you're

59:54

quite right. Doubt and an off switch. That's

59:57

the two things I want the most for machines. But

59:59

I would not even argue that

1:00:01

human beings hallucinate more than these AIs

1:00:03

do. Well, that's, I mean, you

1:00:05

know, Anil Seth and others have talked

1:00:07

about our consciousness as a controlled hallucination. We

1:00:10

do not have direct access to reality. Consciousness

1:00:13

and conscious perception revolve for utility,

1:00:15

not accuracy. So we are coming

1:00:17

up with actionable partial apprehensions of

1:00:19

the world on the basis of

1:00:21

feedback loops that weave in memory

1:00:23

and experience, we confabulate all the

1:00:25

time. Yeah, absolutely. But

1:00:27

interestingly, of course, what we do and

1:00:29

LLMs are not very good at

1:00:31

doing is we have the constant, as

1:00:33

it were, dampening effect of sensory

1:00:36

input that constantly encouraging us to reorient

1:00:38

our hallucination. So our hallucination is

1:00:40

anchored, so to speak. And the question

1:00:42

is, what is a similar anchoring

1:00:44

mechanism for LLMs? Love that. Tom, tell

1:00:46

me the one thing that made you think differently. I

1:00:49

mean, it's a very,

1:00:51

perhaps, obvious answer, but

1:00:53

becoming a parent. No,

1:00:56

this isn't a parent. We have a common

1:00:58

theme here, yes, for sure. Yeah, my children

1:01:00

are 9 -11, and it changes everything. It

1:01:02

doesn't mean everyone should have kids. It's maybe

1:01:04

much more sympathetic to people who don't have

1:01:06

kids. Having kids is very serious business, you

1:01:08

really shouldn't do it unless you really want

1:01:10

them. But just these little

1:01:12

pieces of your heart walking around in

1:01:14

the world and also just the daily

1:01:16

relationship with the extraordinary limitations of your

1:01:18

ability to control things. I can't keep

1:01:20

them safe. I can't make them do

1:01:23

what I want. I can try. I

1:01:25

have to try and create the conditions

1:01:27

that will allow them to become the

1:01:29

best versions of them. And it's an

1:01:31

extraordinarily painful and strange and wonderful thing.

1:01:33

And it, you know, maybe it's teaching

1:01:35

me. how to die, which is the

1:01:37

great task of life. You learn how

1:01:39

to do that well. And

1:01:41

you didn't even mention hallucinations, which is

1:01:43

something they do magically well as well. The

1:01:45

new book is called Wise Animals. There's

1:01:47

also How to Thrive in the Digital Age.

1:01:50

This is Gamora and other fascinating

1:01:52

books. Tom Light people know where

1:01:54

they can find out more about the new

1:01:56

book Wild Animals and where else you're

1:01:58

creating content and what you're up to. So

1:02:00

you will find me online on my

1:02:02

website, Amazon and other bookshops. I also design

1:02:04

Business courses, people like the Economist around

1:02:06

AI and critical thinking. I learned to be

1:02:08

found writing online on Substack and appearing

1:02:10

cheerily on YouTube and other places, talking and

1:02:12

broadcasting. So I'm quite easy to find,

1:02:15

especially if you ask an AI to help

1:02:17

you. I love that. Well, Tom, thanks

1:02:19

so much for your time. It's been

1:02:21

my very great pleasure, Rach. Thank you for having me. Spells

1:02:37

on the air See

1:02:39

there it's crushing The

1:02:41

final impression Stains on

1:02:43

the bay But where

1:02:45

words fell like water

1:02:47

On earth all the

1:02:49

changes That never didn't

1:02:51

matter I think it's

1:02:53

beginning to freeze here

1:02:55

I'm caught in the

1:02:57

rage And the fire

1:02:59

of things All the

1:03:01

brightness that burns me

1:03:03

I'm fumbling through like

1:03:05

a child in the

1:03:07

dark when the nakedness

1:03:09

comes I am shocked

1:03:11

by the color the

1:03:13

glorious way your skin

1:03:15

comes alive and I

1:03:17

never thought we'd make

1:03:19

it back so soon

1:03:21

might be nice Please

1:03:50

forgive me. Could we escape

1:03:52

all the bitterness? Powdered palm

1:03:55

bitterness. Held in the face

1:03:57

of the things that I

1:03:59

don't understand. Intellectual

1:04:01

lies. Over and

1:04:03

over this helplessness suits us.

1:04:06

Funny how quiet has

1:04:08

slipped to our corners. One

1:04:10

of our riches away,

1:04:12

you'll watch it. I'm

1:04:14

breathing and baiting. Wanting

1:04:17

and warming and consciously

1:04:19

weighing. And for some

1:04:21

simple signal To reap

1:04:23

cross your conscience Uncover

1:04:25

redemption And on it

1:04:27

I'll mention I carried

1:04:30

you down To the

1:04:32

St. Lawrence River The

1:04:34

banks running dirty The

1:04:36

waters beginning to freeze

1:04:38

air Solid by morning

1:04:40

And I'll freeze air

1:04:42

This winter by morning

1:04:55

Like so soon, might

1:04:57

be nice But

1:04:59

I knew you'd be

1:05:01

your own destroyer,

1:05:03

cause it's island So

1:05:24

when you'll face such

1:05:26

a curious greatest I

1:05:29

let go your hand

1:05:31

Was desperate to hold

1:05:33

you again But you're

1:05:35

second to deep in

1:05:37

the water Smarted by

1:05:40

myself and so easily

1:05:42

gave up what I

1:05:44

wanted Solid by morning

1:05:46

But I wanted It's

1:05:48

winter by morning back

1:06:00

so soon,

1:06:03

might be

1:06:05

nice But

1:06:08

I knew

1:06:10

you'd be

1:06:13

your own

1:06:15

destroyer And

1:06:20

And I'll freeze

1:06:23

here It's when

1:06:25

I'm by morning