Latest / Cutting Through with Natasha Plowman / Cutting Through with Tim Cook

Transcript

Cutting Through with Tim Cook · Cutting Through with Natasha Plowman

0:14
Welcome to Cutting Through with Me, Natasha Ploughman.
0:17
Today I am joined by Tim Cook, who is the founder and CEO of an
0:22
organisation called AI Confident.
0:24
After many years in the public service and being AI lead for
0:28
government organisations, he set up AI Confident to help people
0:32
navigate the changes that we're seeing in society and business.
0:37
So welcome to cutting through Tim.
0:39
Give me a little bit of a whistle stop to to who you are,
0:42
what makes you tick and what you passionate about when it comes
0:46
to the work that we're talking about today.
0:49
Hi, Sasha, Thank you very much for for having me.
0:51
I'm incredibly passionate about the fact that there are leaders
0:55
across the economy in the private sector, in the third
0:59
sector, in the education sector, who are and in the public sector
1:03
who are concerned about what is coming down the line at them
1:07
from a perspective of AI. So often we hear people talk
1:11
about AI and they immediately go to, you know, impact on the
1:14
bottom line. How can we create efficiencies?
1:16
How can we create cuts? But actually the widespread use
1:20
of AI technology and having so many fundamental changes to
1:24
society and to the economy. And I'm passionate about helping
1:28
leaders see that coming and and be ready and prepared for the
1:32
much more strategic change that's coming as a result of AI
1:36
technologies. And see this as much more than
1:38
just how do we help our staff to use Co pilot or ChatGPT.
1:42
But actually I need to reshape my organisation, my business or
1:45
what we care about because in my view this technology is going to
1:49
have a fundamental change across the economy and across society
1:53
and as many leaders as possible who are ready for that is what
1:57
drives me, gets me out of bed. Brilliant.
2:00
So let's take a step back and start talking about what is
2:04
artificial intelligence and what is AI.
2:06
It seems to be an all encompassing term for a lot of
2:08
different types of technologies. So what would be your go to
2:12
description of what we're talking about when we talk about
2:15
AI? Let's say this is a often talked
2:17
about and something I try and avoid as much as possible.
2:20
One of my jobs, previous jobs in government was to write the
2:23
first legal definition in UK law of artificial intelligence,
2:27
which is still there. But I think the definition is
2:31
less important than having an understanding of the range of
2:37
different AI technologies. The fact that artificial
2:39
different technologies, some of which have been with us for a
2:42
long time, many of which we've been using in our day to day
2:46
over and over again. And only some of which fall into
2:50
the category of generative artificial intelligence, which
2:53
is what's causing all this stare with GPT and everything else.
2:57
Yeah, and I think that's The thing is that when people talk
3:00
about AI, they they are talking about the things that have
3:02
really blown into people's consciousness around large
3:05
language models and ChatGPT. But I'm really interested in
3:09
that bit that what is artificial intelligence because and how
3:12
it's been used for decades, that people may not realise that
3:15
they've been understanding that this is also within the banner
3:18
of AI as well. Absolutely.
3:21
Lots of things that we do on a day to day basis are AI driven
3:25
from for example, if you are using something like Netflix or
3:30
the content recommendation that sits in your social media feed.
3:34
Lots of what we call traditional machine learning technology,
3:37
which is a technology that's been around with us for quite
3:40
some time and is the the premise of machines effectively learning
3:43
from the interactions that they have with us as humans.
3:46
They have a particular goal to achieve and from what they learn
3:50
from their interactions with us, they change up future
3:53
interactions with us in order to achieve that goal.
3:55
To make that really clear, if you take something like Netflix,
3:59
for example, it has the goal of getting you to watch more
4:02
content on Netflix. And it does that by learning
4:04
from what you've watched previously and showing you
4:06
similar content in the future. And similar things are happening
4:09
with your social media recommendation feeds and so on.
4:12
And this machine learning technology has been around with
4:14
us for quite some time. It's pretty robust, pretty well
4:17
known about. It's doing things like when we
4:20
start typing in Google, giving us that prediction of what the
4:22
next word is that we're going to going to type all of that kind
4:26
of stuff that we've seen around with us for quite some time.
4:29
There's then technology or branch of machine learning,
4:32
which is called deep learning. And this has probably been with
4:34
us since about the 20 tens and is where artificial intelligence
4:38
starts to act a little bit more like the human brain.
4:41
And we start to see capabilities like the ability to do facial
4:44
recognition, the ability to differentiate between a cat and
4:47
a dog in a in an image. And so these kinds of deep
4:51
learning technologies and say been with us, it's about 2010 is
4:53
also enabling us to do stuff like spot tumours in images of
4:57
brains, spot weapons exposes in X ray, images of bags.
5:01
All of these things have been developed for a while.
5:03
And again, you're using these when you open your phone using
5:07
biometrics or when you do a reverse image search on Google
5:10
or whatever. Again, technologies that have
5:12
been around with us for some time.
5:14
And the important thing about AI technologies like machine
5:17
learning and deep learning is that they're largely doing two
5:20
things. One of those two things is
5:21
they're categorising, right? The AI technology is
5:24
categorising something as something you want to watch or
5:26
something you don't want to watch.
5:27
It's your face. It's not your face.
5:29
And secondly, then narrow by narrow, if you talk about the AI
5:34
technology that is helping spot images of tumours and brains, it
5:38
is better than human level capable at that one task.
5:41
Rubbish anything else, but better than human level capable
5:44
at that one task. What we've seen with generative
5:47
AI is a shift of both of those paradigms.
5:49
So generative AI moves from just being able to classify things to
5:53
AI technologies that are able to generate new content and
5:56
generating that new content in imagery and video, in text, in
6:00
voice, in podcasts, in all sorts of different ways.
6:03
So it's that shift. First of all, the shift from
6:05
being able to to categorise things to being able to generate
6:09
new content. And secondly, we're moving from
6:12
AI technologies that are really narrow in their capability of
6:15
their ability to do 1 task, two generative AI technologies that
6:19
are at or near human level, capable in an incredibly broad
6:23
range of tasks and able to solve millions of different types of
6:26
problems for individuals as we go and use them.
6:29
We all have very different use cases.
6:31
And when we go in and speak to organisations, one of the first
6:34
things we do is get everybody to put on the table their different
6:36
use cases. And they're so broad and so
6:39
varied using the same general purpose generative AI tool.
6:43
And those two shifts fundamental in Y of AI technology has come
6:49
much more into the public domain and, and, or if you like, beyond
6:53
the other AI technology that have been around for a while.
6:56
But it's also important to remember that those AI
6:58
technologies that we know a lot about, that have been around for
7:00
a while are sometimes the best way to achieve a task, and not
7:04
always is generous of AI the right approach to use for a
7:08
particular task. A big thing I want to talk about
7:11
is how we seem to get captured by the new shiny, but still
7:14
don't really understand the foundations or the fundamentals
7:18
of some of the technology we're using.
7:20
Society is likened to a similar technological revolution, like
7:24
the industrial revolution in the late last century.
7:27
Do you feel that we're in a similar seismic changes to how
7:31
society is ordered and operated? And do you have an idea of what
7:35
point of that stage that we might be at the moment?
7:40
Yes, I definitely do think this is going to change a lot.
7:43
I think we're still very early days in that evolution.
7:46
I think one of the big differences this time around is
7:50
that these technologies are coming for knowledge work.
7:55
So previously with previous industrial revolutions, it's
7:59
been manual labour that has been been automated.
8:02
Out here we have the technology that is that is coming for
8:04
knowledge work and, and in many cases creative work as well.
8:08
I think the pace of that change is quite varied across the
8:13
economy. There are some jobs and job
8:16
functions that are already, we've seen quite a bit of
8:19
disruption in IO talking to copywriters for example, or
8:22
people in certain creative industries where we're already
8:26
seeing quite a lot of disruption.
8:28
I think there is quite a lot of disruption coming down the line
8:34
for large bureaucratic organisations, but they are
8:37
going to be slower to for that kind of adoption to philtre
8:40
through within the organisation. But everybody needs to be having
8:44
a look at their job role and their job function and having a
8:48
think about how that might change in a world where we have
8:54
a lot of AI capability sat around us as humans.
8:57
And I often talk about the the ability to have you as the
9:02
expert plus AI is always going to be better than the expert on
9:06
their own or the AI on its own. And having a think about what
9:10
that looks like for your particular job is going to be
9:12
really important. And it is that whole area of
9:15
where does AI stop and start, where does human stop and start?
9:18
And what differentiates human ability, knowledge, nuance
9:24
versus AI. And you've got some schools of
9:27
thought where they just seem to be going so far down the path of
9:30
just saying humanity isn't required in the same way and
9:33
we're going to have no role for society, which obviously evokes
9:37
fear in a lot of people. And we can't imagine that world,
9:41
except for some of the sci fi that we might be watching.
9:43
The optimist in me, which I really want to hold on to in the
9:47
other times where you feel a bit negative about where we're at it
9:49
as a society is that there is so much more to come that will make
9:54
society a little bit more ordered, organised and there's
9:57
still that role of humanity. So where do you sit on this
10:01
optimist versus pessimist or realist kind of spectrum?
10:05
I change every moment of every day, to be honest with you and
10:08
Tasha, I go through the full gamut of emotions over the
10:10
course of most days depending on what I've read, what I've
10:13
listened to or who I've spoken to.
10:15
But listen, I think there are a couple of things that are
10:17
important. And you talked just a moment ago
10:20
about you alluded to doing more as an organisation.
10:25
And I think that's going to, there are two kind of
10:27
fundamental thought processes that organisations and leaders
10:30
and individuals are going to have to go through.
10:32
The 1st is what do we want to do more of?
10:35
Because AI technologies allow us to do more.
10:37
And if our leadership approach to bring AI technologies into an
10:41
organisation is our level of ambition is going to stay the
10:43
same, but we're going to bring AI technologies into the
10:46
organisation. That is the obvious outcome from
10:50
that is we're going to need fewer people to do the work.
10:53
But if your leadership position instead is with the people that
10:57
we've got, plus the suite of AI technologies, what more could we
11:02
achieve? You are all of a sudden changing
11:05
the ambition, you're changing the the kind of communication
11:09
with your people. No longer is this about a
11:11
reductive, but it's actually about what more could we achieve
11:14
as an organisation. And then we've got to ask you a
11:17
question of where can we achieve more value?
11:19
And is that through doing more of the same thing?
11:22
So where we were able to do five things for a client in a week,
11:25
we can now do 10, we can now do 15.
11:27
So we can increase the number of clients that we can serve.
11:29
Or does it mean we can do something with more quality so
11:32
we can improve the quality of the product that we provide and
11:35
we can charge a higher a higher fee for it and do more of that
11:39
kind of work? Or does it allow us to serve
11:42
more people, so to turn say a nine to five service into a 24/7
11:46
service, or to take a service that we deliver and turn it into
11:50
a product and reach more people with that product than we ever
11:54
could with the service that we have?
11:56
Or does it enable us to do something different altogether?
11:59
And I've worked with organisations who have gone
12:01
actually the time that we saved using AI over here means our
12:04
staff can do more CPD over here and improve themselves as
12:07
individuals and grow and develop.
12:09
I've worked with organisations whose staff have said we want to
12:12
go to a four day working week. And the challenge back from
12:15
leadership has been, well, you bring AI tools into the
12:18
organisation, show us you can achieve the same level of output
12:21
and we can, you can have that time back in, in your personal,
12:24
in your family time. So there were different kind of
12:27
ways of thinking about how we do more with AI technologies.
12:30
And then the second thing, which comes back to what you started
12:33
off your question with, is asking the question of what we
12:37
do as an organisation that is uniquely human.
12:41
What are the new uniquely human things that we do as an
12:44
organisation? What do our people enjoy showing
12:46
up to work to do? What are the things that our
12:48
clients value being a human to human interaction?
12:53
And let's put that at the heart of our AI use and go.
12:57
Those are the things that would enable us to do more of.
13:01
And those are definitely the things that we're going to make
13:04
sure we don't automate out by accident as we talk about how we
13:08
use AI technologies. One of the other people that
13:10
I've had in the podcast, Louisa Loran, and she has worked at
13:14
Google, worked at Mersk, and that was a really interesting
13:17
insight. Too often we syphon off the
13:21
things that seem to be easy rather than this, and rather
13:23
than focus on syphoning off the things that are just not
13:26
required for humans to be involved in.
13:29
And really the hard introspective look at your
13:33
business and being really clear about what makes you unique is
13:36
the area that you should ring fence in a way and make sure
13:40
that it's really core to what the business moves and does into
13:43
the future. And and then you have this other
13:45
conversation. I think this is where we go back
13:47
to the whole point of how do we cut through so much of the just
13:52
just the, the fear, which is making people worried about the
13:55
future and careers and what society is going to look like
13:59
and gives us some clearer path. Or is that just something that
14:04
we just need to accept isn't going to happen because things
14:05
are changing so much that there isn't really a clear path of
14:08
where development might go? There's two different parts to
14:11
that question. Is there a clear path to where
14:13
this is all going to go? No, I don't think there is.
14:15
But what I think that we absolutely do have at the moment
14:18
is agency. And individuals and
14:20
organisations have agency over the path that they want to take
14:23
with these technologies, how they want to bring them into the
14:26
organisation. And we mustn't lose, we mustn't
14:29
feel like we don't have agency over that.
14:31
And so for leaders, it's kind of two prongs to that.
14:34
One of those is how do we give the right leadership messages to
14:39
our staff? Yeah.
14:41
To make them feel like they have agency and to encourage that
14:45
kind of agency within the new organisation and to take
14:48
proactive decisions ourselves as leaders about all of those, the
14:51
kind of things that we talked about there.
14:53
But I think one of the most powerful things that you can do
14:56
if you're leading an organisation or in a team is
14:59
just to open up a conversation with your staff.
15:01
How do you feel about AI technologies?
15:03
What is it that is concerning you at the moment?
15:06
And because everybody will have something that they're
15:11
harbouring or that they're worried about and just getting
15:13
that conversation out in the open, first of all, and getting
15:16
some different challenge on those few points from different
15:18
people in the team. As a leader, it's really
15:20
difficult to open that conversation because you don't
15:22
have all the answers and nobody has all the answers at the
15:24
moment. But being able to lead those
15:26
conversations from a place of vulnerability, from a, a place
15:29
of not having all the answers, but just helping your team
15:32
surface those conversations is the first kind of entry point
15:36
to, to building that, that confidence.
15:39
And the second thing I would add to that, I, I always say there's
15:42
two things that you shouldn't pay us to come and do for you
15:45
because you can do them yourselves.
15:46
One of them is ask your staff how they're feeling about AI
15:49
technologies. The other is ask your staff how
15:51
they're using AI technology at the moment and just surface some
15:55
of that and create that really transparent culture of sharing
15:58
of that environment around around these tools.
16:01
Because that will be a first step in cutting through some of
16:05
that reticence. Because every organisation has a
16:10
person, at least one person who is utterly terrified of these
16:14
technologies, won't go near them and feels like the business is
16:19
moving too fast. While at the same time in that
16:22
same organisation is the person who has completely engaged with
16:26
these technologies and feels like the business isn't moving
16:28
fast enough. And as leaders, we have to
16:30
manage both of those perspectives and bring those
16:32
people in the same room. And often those the assumptions
16:36
is that's around age and it's not always the case, is it this
16:39
whole idea? And, and I'm really intrigued
16:41
when we there's a whole slight segue here, but a whole
16:44
conversation around people over 50 not being employed, been seen
16:48
as Luddites, not seen as grasping technology.
16:51
But the people who I've come across who are older, more
16:54
experienced, have more depth and knowledge, the ones who really
16:57
understand the technology a lot more than what the assumptions
16:59
are we have about young people. So is there something that we
17:02
need? So it's interesting you talk
17:03
about, we need to open the conversation with what are your
17:07
thoughts, feelings about the technology that you might and
17:10
what and how you're using the technology as well, rather than
17:13
just assumption that someone's over 50, they don't know how to
17:16
use any AI tools, which is obviously not true at all.
17:21
So what would be really interesting then is
17:25
understanding and unpicking people's fears.
17:28
So what are you hearing the main things that people tend to be
17:32
most fearful of? And how are you?
17:34
Obviously your name of your business, AI Confident is core
17:37
to this. So what are you hearing as the
17:39
main things that people are fearful of?
17:41
One of them is definitely lack of clarity over what they can
17:45
and can't put into the AI tool from a data perspective, and
17:50
that is particularly prevalent in organisations who don't have
17:53
clear guidance for staff on where their data or how they can
17:58
and can't use data with AI technologies.
18:00
Then there's definitely people who have a, a dislike that they
18:06
have harboured about AI technologies from a particular
18:09
angle, which is all very valid and understandable, but at the
18:12
same time as prevented them then engaging with the technology at
18:15
all. And the fact that they are
18:18
concerned over here has meant that they haven't taken any of
18:21
the opportunity that exists. And, and I think that's a real,
18:25
you know, helping people just to see from their peers how they
18:29
are using it. Yes, you may have very valid
18:31
concerns about the energy consumption of generative AI
18:34
technologies or you might have very valid concerns about how
18:38
these tools have been trained or the, the material that they've
18:41
been trained on or whatever. But if we leave all the value on
18:44
the table, we're only going to accumulate that risk and those
18:50
downsides of the tools. And so I very much come from a
18:55
viewpoint of we've got to take as much of the opportunity from
18:59
these tools as we possibly can while also doing so in full
19:03
cognizance of the risks, challenges, issues that there
19:06
are with them and driving those down and challenging those as
19:10
much as we possibly can. And I want to.
19:11
Talk a bit more about some of the risks.
19:13
Obviously you came from government and we're looking at
19:15
the regulations around it. And it's very hard for nation
19:18
States and individual governments to regulate these
19:21
kind of technologies, which cross borders inherently.
19:24
And also issues around privacy and trust and misinformation
19:29
corresponding with the rise of the use of generative AI.
19:32
We're seeing an explosion of misinformation and
19:35
disinformation where you don't know what's real.
19:38
You used to trust imagery and video because you could see that
19:41
was living and breathing and you could see the people behind it,
19:43
but now you can't. We used to be able to trust
19:46
someone speaking and now we can't.
19:49
So is do you think this this real fundamental distrust of
19:53
some technology is a stage, or do you think that's, is there
19:57
more we need to be doing in society to prepare people and to
20:01
give them those tools to be able to see what's real and what's
20:04
not? And trust is a really important
20:06
thing here. Just very personal and niche
20:09
example. I missed a really important
20:12
email from HMRC the other day because it was the 5th email
20:16
purporting to be from HMRC that I'd received in the week and the
20:19
others of which were all complete phishing attempts.
20:21
And so I just think of this next one as well.
20:23
And, but don't worry, I'm on top of my taxi if anyone if anyone's
20:26
asking. But but that's a classic example
20:29
of how a lack of trust all of a sudden because of the volume of
20:33
misinformation, it is inhibiting people's decision making in
20:37
their lives. To be honest, I think I'm not
20:40
convinced it's a technological solution them here.
20:42
And because the moment you find a technological solution,
20:46
there'll be somebody else who develops the technology that
20:50
defeats that particular solution.
20:52
And I do a lot of work with schools and I do a lot of work
20:55
with charity funders who are in different ways concerned about
20:59
the other people's use of AI technologies to gain their
21:03
assessment approach in my shape or form.
21:06
And all the while there are people who are developing
21:08
technologies that purport to spot AI use in those processes.
21:13
But all the while someone's investing in that.
21:16
There's somebody else who's investing in the technology that
21:18
will defeat the AI detection technology.
21:21
And, and I fear that's a a kind of money pit down which we don't
21:26
want to Chuck loads of our hard earned resources.
21:30
And so we have to find different kind of societal approaches to
21:35
trust and managing information and understanding and verifying
21:38
the the rest of information, which includes a good degree of
21:43
literacy around some of the technological, some around AI
21:48
technologies, how they impact our lives.
21:51
How people can take algorithmic based or how sorry, AI tools
21:55
take algorithmic based decisions around what they show to you and
21:58
how that can influence influence your kind of trust or your
22:01
perception on different, different things.
22:03
I think there is a really low level of literacy around things
22:07
like data security and data protection around the how some
22:11
of the algorithms work. The fact that I can be sat next
22:14
to my wife looking at the same piece of content on social media
22:18
and we'll see different comments prioritised to the top because
22:21
because of our individual references, which then comes to
22:25
a people need to actively know how to need to actively seek out
22:32
counteracting viewpoints to the ones that they're being being
22:37
shown. And there's a whole set of very
22:38
human literacy tools that we need in that space that are set
22:43
from a thinking that this is just a technology, There's some
22:45
technological solution to this and we don't know.
22:48
What are you doing around particularly young people in
22:51
schools and helping schools prepare our young people for a
22:54
future? Because for me, that's the one
22:57
thing I'm most concerned about that we're not.
22:59
We're not having enough of a debate and conversation about
23:02
what's real and what's not, who to trust and who not to trust,
23:05
because we're just bombarded with noise.
23:07
And I think that's again a premise of what cutting through
23:09
is, just the noise and you can't decipher what is actually real
23:13
or not. So do you think it starts with
23:15
schools or? It does, and there's definitely
23:17
a piece around how we help young people to be critical consumers
23:21
of of dutiful media, but it's much broader than that.
23:24
For schools, the issue extends beyond cheating in exams or
23:31
misinformation. Disinformation.
23:33
To be frank, the thing that concerns me more than anything
23:36
else is the use of companionship type chat bots by people under
23:42
the age of 18. Particularly where those
23:44
companionship chat bots are designed not with the users best
23:47
interests at heart, but they're designed for commercial monetary
23:51
and they are designed to keep the young person interacting
23:54
with that tool. We know that AI tools can be
23:58
incredibly sycophantic, can be incredibly engaging, can sound
24:01
incredibly human like and can take people down all the wrong
24:05
places and not provide good challenge and give them really
24:09
warped perceptions of what good relationships look like.
24:13
And when you're when you're a child and you have spent a lot
24:18
of time engaging with AAI companion who's never said no to
24:22
you, who's never disagreed with you, he's gone along with
24:24
everything that you've said. What perception does that give
24:27
young people of what good, healthy relationships look like?
24:30
Where we do have disagreements, we do have to work through
24:33
differing perspectives on things.
24:35
We do have to deal with challenge to our kind of world
24:38
view and everything else. And not many people are talking.
24:42
Not enough people are talking about that in in my view.
24:45
And that is the thing that Worsley.
24:47
Just the whole notion of having healthy debate, healthy
24:50
conflict, healthy disagreement is something that at all ages
24:54
seems to have been lost either through fear of causing offence.
24:58
But your point around if you live in a world where no one
25:01
says no to you and everyone, everything disagrees with what
25:04
you say, that isn't real world at all.
25:07
And I know for me, one of the things that I quickly did was
25:10
turn off AI, the AI generated chatbot on my all of my
25:14
daughter's social media. I think social media has pluses
25:18
as well as minuses. But yeah, I think I was hearing
25:20
from you is one of those things that I cut off very quickly and
25:23
have not reinstated, But thank you for that.
25:26
But look, this has been a really interesting conversation.
25:29
As with anything to do with AI, it covers so many different
25:33
areas and possibilities of what people can do.
25:36
I think the main thing is from what you've said is we need to
25:39
be a lot more open, objective and honest around what tools
25:43
we're using, what our fears are, how we can approach the future
25:46
in a different way and with positivity, as well as openness
25:51
to where the negatives may lie. So this is a really fascinating
25:55
conversation and thank you very much for your time.
25:59
Thank you for the great questions and Sasha, I really
26:00
enjoyed it. Thank you.