0:00

This is the BBC. This

0:03

podcast is supported by advertising

0:05

outside the UK.

0:07

BBC Sounds,

0:10

music, radio, podcasts. Welcome

0:18

to Understand Tech and AI,

0:21

the podcast that takes you back to basics

0:24

to explain, explore, unpick

0:26

and demystify the technology that's becoming part

0:29

of our everyday lives. I'm

0:31

Spencer Kelly from BBC Click and

0:33

you can find all of these episodes on

0:35

BBC Sounds.

0:42

Throughout history we have toiled

0:45

with works, we've built things,

0:48

ploughed things, grown And

0:50

this is how we've earned a wage

0:52

to feed and house and

0:54

clothe ourselves. And that's been fine, thank

0:57

you very much. But also throughout

0:59

history, we've learned how to build machines

1:02

that are faster, cheaper and

1:05

more efficient than people. Machines that

1:07

can do more of the repetitive, physical, dangerous

1:10

manual labour that we'd really rather not

1:12

do anyway. There have been many

1:14

ages of automation through the centuries

1:17

and now we are facing a new one,

1:20

the age of artificial intelligence.

1:24

And this time it's not the physical work

1:26

that's being automated, it's the

1:28

thinking work. It's work

1:30

that requires conversation and discourse

1:32

with others, work that's traditionally

1:35

felt pretty human. It's

1:37

everything. Is AI

1:40

coming for your job?

1:41

Is there any type of work that's safe from

1:43

an AI takeover? Well

1:46

while I've still got a job, I'm joined by Dr

1:48

Jonathan Aitken. You've still got a job as well, yeah? I

1:50

do, yes. Brilliant. You are an electronic

1:52

engineer, doctor of applied AI and you

1:54

teach robotics at Sheffield University.

1:57

Absolutely. So you are well placed to

1:59

talk about...

1:59

the jobs that AI may or may not take.

2:02

Give us an overarching view first of all, how

2:05

worried should we be that AI is

2:07

going to replace a lot of our jobs?

2:09

As ever through every industrial revolution

2:12

that we've seen, we're going to see a

2:14

change in the nature of jobs. Whenever

2:17

I answer this question, my first thought

2:19

is always to think back through the trajectory of work

2:22

and see how it's been affected by each one of these,

2:24

whether it be mechanisation, electrification,

2:27

computerisation. And right throughout

2:29

that history, we've always seen an increasing number

2:32

of jobs because wherever we've

2:34

applied a new technology, new jobs

2:36

have been created. Yes, we have

2:38

seen a shift in the working patterns, we've seen

2:40

a change, but ultimately the number

2:42

of people in employment has kept growing

2:45

right throughout that time. Is that because

2:47

we're creating jobs that

2:49

we didn't know we needed because of the

2:52

new technology, or are we just more

2:54

affluent so new industries open up

2:56

because we don't all have to go down the mines seven

2:59

days a week? I think it's a little bit of both.

3:02

I think absolutely we're going to see as new

3:05

technology evolves, we see

3:07

new opportunities and we'll see things change.

3:10

So if we look to kind of the previous industrial

3:12

revolutions, we'll see more supervisory

3:14

roles appear where people begin to look after the hardware.

3:17

And actually, when you introduce the hardware,

3:19

you

3:20

still need something to service the hardware, you still need something

3:22

to maintain it, you still need something to work on it. So ultimately,

3:25

we see the nature of some of those jobs start

3:27

to change, they can either become more managerial, or

3:29

they can become more practical depending

3:31

on what routes people would want to take. As

3:34

you say, automation is a fact of life. And

3:36

throughout history, we built machines that

3:38

replaced manual labour. I've heard

3:40

it said quite a lot now, that

3:43

this time it feels different

3:45

because we're not replacing physical

3:47

labour. And we're all going into kind of

3:49

thinky jobs,

3:50

intellectual jobs, what we're seeing

3:52

is AI being able to do at least

3:54

some part of those thinky

3:57

intellectual jobs. So it begs the question, where do

3:59

we go?

3:59

if it's taking that bit

4:02

away? I think it's always

4:04

an interesting question because ultimately

4:07

how does the relationship with the AI work

4:09

as part of the job? Where do the bounds

4:11

sit? Now we're sitting at a point

4:14

in history where we're not quite sure because lots

4:17

of people will make predictions for what we're going to see

4:19

in five years, in ten years, in fifteen

4:21

years but we don't know how the technology

4:24

is going to develop. We don't know how we're

4:26

going to develop with the technology as well. Ultimately

4:30

we can think about actually is AI more of

4:32

an assistant to us because it's

4:34

become one of those tools and it will

4:36

become one of those tools which is very good at

4:39

giving us select information and being

4:41

able to filter down information.

4:43

So in the past when someone would say I'll just

4:46

go Google it, actually

4:48

this is going to be an easy way to be able to filter that

4:50

information to allow us to be more efficient as

4:53

part of our jobs rather than spending 20 minutes

4:55

looking through search answers and clicking on a

4:58

few and kind of going off into

5:00

the ether and potentially be distracted

5:02

in various other elements we can actually now

5:04

get one single answer delivered

5:06

to us in a very very efficient manner

5:08

which hopefully solves our conundrum.

5:11

Okay Jonathan let's take a short break

5:13

because it turns out that

5:15

intelligent machines have been coming for our

5:17

jobs and for us for

5:20

longer than many of us have realized. Our

5:22

tech historian James Sumner picks up

5:24

the story.

5:51

The word robot first appeared in 1921 in the play

5:53

R.U.R.

5:59

where it described a kind of emotionless

6:02

artificial humanoid, mass produced

6:04

in a factory, and set to work on

6:06

menial jobs. Even before

6:09

the robots finally rise up and destroy

6:11

their creators, their existence is shown

6:13

to be destroying human society,

6:16

as most workers lose their jobs, and

6:18

the world economy plunges into chaos.

6:22

Since the Industrial Revolution, these

6:24

fears had focused very much on manual

6:27

jobs, but in the 1950s,

6:29

engineers have always made machines

6:32

that move. Now they are beginning to

6:34

make machines that appear to think. With

6:36

excitable media reports of electronic

6:39

brains, secretaries, administrators,

6:42

and even doctors and lawyers were

6:45

suddenly in the firing line for potential

6:47

replacement. Of course,

6:50

not everyone agreed that machines destroyed

6:52

livelihoods. The work, after

6:54

all, would still be done, and the

6:56

benefits would go to someone. What

6:59

if we simply agreed to share them? And

7:01

computerisation created some jobs

7:04

itself, often more interesting

7:06

than the ones that had been lost. Leon

7:09

Bagret, head of the British computer

7:11

firm Elliot Automation, was

7:13

particularly active in promoting this

7:15

vision. Instead of reshaping

7:18

the availability of work around changes

7:20

in the technology, true automation

7:22

would fit the technology around humans'

7:25

ability and desire to work, as

7:27

he explained

7:27

in 1964. The

7:30

opponents of automation are basically

7:32

people who are pessimists. Somehow

7:35

they don't believe that human beings can be

7:37

trusted with riches and leisure, tell

7:40

them that here's a way in which we can all be

7:42

better off, and they warn us solemnly with

7:44

a wagging finger to be aware of affluence.

7:48

That was Dr James Sumner with

7:50

the first predictions of the rise of the robots.

7:54

Now, Jonathan, what sort

7:56

of jobs do you think might be at

7:58

risk at the end of the world?

7:59

Can you think of any jobs that

8:02

are safe? I think I'll start with the jobs

8:04

that are definitely safe a lot of the practical

8:06

physical jobs a lot of the kind of manual

8:09

Work that we do they're going to be safe for a

8:11

very long period of time So if they haven't been

8:13

replaced by machines so far like making cars if

8:15

they haven't already fallen to the robots

8:18

You're saying they are now safe. The

8:20

world is a complex place Humans

8:22

are very good at dealing with problems. We're

8:24

very good at dealing with issues that occur.

8:27

We're very good at being dexterous We're very

8:29

good at being craftsmen and actually dealing

8:32

with very nuanced tasks A robot is

8:34

still very good at dealing with the one task that you put

8:36

in front of it But as soon as you ask it to walk

8:38

over wobbly ground and pick up something that's

8:40

not quite in the right place and do something New with it a

8:43

robot just goes to pieces basically. Absolutely

8:45

when you work in an uncertain world A robot

8:48

can be more of a hindrance than

8:50

it can be a help

8:51

Now I've spoken previously about

8:54

my hatred of the photos of the terminator

8:56

or the strange white humanoid robot That's attached

8:58

to most news stories about AI Robots

9:01

are dumb mechanical machines that

9:04

do one thing and AI

9:06

is something that Learns and can

9:08

adapt but the two can

9:10

be combined can't they should we be thinking

9:13

about robots that do? Walk

9:15

around our world or will themselves around our world

9:18

and learn and adapt and get better at

9:20

Most things that we could do. I think

9:23

there's always two different questions with AI

9:25

and it's it's almost the the general AI

9:28

Versus a specific use case AI

9:30

where we're looking at AI which is adapted to

9:33

within a particular task So yeah, we used to get

9:35

the AI playing games like go and chess

9:37

and everything else When we talk about the robots

9:40

kind of wandering around the world We're now talking much

9:42

more into the the sense of general AI into

9:44

the sense of AI that we're letting loose

9:46

in the world And that has an

9:49

enormous amount of difficulties Not

9:50

just because it has some kind

9:52

of physical form the robot which is going to struggle

9:55

on Various different types of terrain

9:57

depending on how well it's been built and how much money

9:59

people

9:59

have paid for it. But it

10:02

also has the difficulty of the complexity of

10:04

the world, sensing the world and understanding

10:07

how the world fits together because inherently

10:09

the world is a noisy, difficult

10:13

to understand place with lots of

10:15

minute changes which can actually have

10:18

very big impacts as part of it. And

10:21

AI is not well suited to that because

10:23

the amount of training, the amount of learning

10:26

that the system needs to do in order to be

10:29

able to pick

10:29

up those nuances is

10:32

significant. Okay, what sorts

10:34

of jobs might be at risk then?

10:37

I think the types of jobs we've got to think about or the

10:39

types. I wouldn't say at risk. I think it's,

10:42

I think again, it's

10:43

about thinking about how the nature of the job is going

10:45

to change. And a lot of those could be things

10:48

around decision making. I know we've seen AI

10:50

applied, particularly in health

10:52

services and health sectors, looking at automated

10:55

analysis of things like images and trying to pull

10:57

out elements and try and pull out pieces.

10:59

Try and spot tumors, for example, in x-rays.

11:02

But as people, we

11:04

always want a human sitting

11:06

there on the side of it, checking over

11:08

it and actually making sure that we're

11:10

happy with that decision. A question

11:12

I'd often ask people around this is to say, well,

11:15

let's say we're designing and building a new aircraft. I'll

11:17

have an AI algorithm

11:19

do all the programming for it and create all the software

11:21

for it, build it all, program it all, encode

11:24

it, stick it on the computers, stick it

11:27

in the aircraft, and then we'll fly it. Will

11:29

you get on board that aircraft? No,

11:32

you can go first. Exactly. But that's

11:34

the key point because we are still people. So

11:37

whilst we think about how AI

11:39

is going to have an impact, and it will have an impact, we

11:41

as people still want people involved. Here's

11:44

another issue.

11:45

When you employ humans, you pay them a wage

11:47

and those humans pay part of that wage as income

11:49

tax. When you replace those with

11:51

machines, you don't pay those machines a wage,

11:54

those machines do not pay income tax, and

11:56

the coffers of the country get depleted. What

11:59

is the solution?

11:59

to that, who

12:02

pays for the roads in the hospitals if you've got no

12:04

human workers earning wages? I

12:06

think this becomes one of those things where

12:08

we may need to discuss issues like a basic

12:10

universal income depending on depending

12:13

on how far we go with these processes because

12:16

again there is always a choice for this there is always

12:18

an option whether how much we implement

12:20

some of these elements and I think

12:23

you're absolutely right asking the question is that then there

12:25

also needs to be a plan there that says if

12:27

we do this and actually if we do it

12:29

in a very aggressive manner where

12:32

actually we do begin to remove

12:34

big chunks of the workforce actually how

12:36

is that element of the workforce going to live and

12:39

how actually is the economy going to change

12:41

as well as part of that? One solution

12:43

I've heard is a robot tax where if

12:45

you have a robot doing a job then you pay something

12:48

equivalent to income tax you

12:50

know the company that makes it contributes to the country

12:53

in the same way that a human earning

12:55

a wage might. I think it's an excellent idea

12:57

and I definitely think it's something that we can consider. I

12:59

think not at the moment particularly because

13:02

we're still in very early days

13:04

for a lot of the industry and actually we

13:06

still don't know exactly where it will go particularly in

13:08

some parts of the processes but it's

13:11

the kind of thing that we may need to think about actually

13:14

if we do see significant changes.

13:17

Jonathan thanks so much for your time thanks

13:19

for helping us understand what work

13:22

might look like in the future. Thank

13:24

you very much it's been a pleasure. Now

13:26

as well as worrying about our jobs there

13:28

is one other big

13:29

question that comes up whenever AI

13:32

is mentioned thanks to all

13:34

of the sci-fi films ever we

13:36

are very worried about what happens

13:39

if AI becomes too powerful.

13:42

Will we be able to control it

13:45

or could it stop us from switching

13:48

it off?

13:49

That is what we'll be talking about in the final

13:52

part of this series. Do

13:54

join us your survival

13:56

may depend on it.

13:59

I'm Helena Bonham

14:01

Carter and for BBC Radio 4

14:04

this is History's Secret Heroes. She

14:07

received

14:08

a brown envelope and says do not

14:11

open it until you get on the plane. A

14:13

series of rarely heard tales from

14:15

World War Two. They knew they were going to be

14:17

caught and actually that was sort

14:19

of part of the plan. Unsung

14:22

heroes, acts of resistance, deception

14:25

and courage.

14:25

That is a morning that is seared

14:28

into my memory. I will

14:30

never be able to forget the terror

14:33

of that morning.