Business and Trade Committee

Wednesday 8th July 2026

(began 2 months, 3 weeks ago)

Share Debate
Watch Live
Skip to latest contribution
This debate has concluded
14:30
None Portrait Speaker 1
-

And other areas like coding. And those are mostly tasks which are verifiable. In terms of impact of AI on labor markets, we're seeing some impact in certain pockets like translation work. And I think one of the key reasons that we will still need humans in the loop for some time going forward is that most activities still need human verification. And so AI makes mistakes and we need workers that are capable of verifying the output. That doesn't mean that there is not going to be an impact on labor markets going forward.

And I think one critical question is where are those workers going to be located?And so, for example, if AI makes accounting much easier and it. Is to reason that some of those jobs might migrate to places where labor costs are cheaper, for example. So we may see more accountants being hired in Manila rather than in the City of London. And so AI reducing barriers to entry in knowledge works and professional services, I think, is one of the key mechanisms by which we will see an impact of AI on the labor market relatively soon.

And unfortunately, we lack the data to track these trends in real time. And another mechanisms, which I think is underrated, but we're beginning to see in the data as well, is that more and more people are using AI to fix various things themselves. And so if I use AI, for example, to fix my boiler, that may not replace the heat engineer, but it means that the heat engineer will lose a call and lose one job. Right?And so I think this working assumption is that AI is mainly going to have an impact on knowledge work and professional services.

But I think as more people do more things themselves, we are going to have an impact on manual trades, um, as well. And we will see more firms actually offload verification to their consumers, uh, because that is just, you know, um, easier to do, uh, further down the line. I do think we will see some outright automation, uh, likely in transportation logistics that often takes more time because you need restructuring. And so often when you look at some of these studies that count up to tasks and say that, you know, 70% further down the line, I do think we will see some outright automation, uh, likely in to the consumers, uh, because that is just, you know, easier to do exportation, logistics that often takes more time because you need restructuring.

And so often when you look at some of these studies that count up the tasks and say that, you know, 70% of the job is automatable, but the remaining 25% is not. Well, the question is whether those tasks are needed to begin to the consumers, uh, because that is just, you know, um, easier to do, uh, further down the line. I do think we will see some outright automation, uh, likely in transportation logistics. That often takes more time because you need restructuring. And so often when you look at some of these studies that count up the tasks and say that, you know, 70% of the job is automatable, but the remaining 25% is not.

Well, the question is whether those tasks are needed to begin with. And often the way that automation proceeds is by simplifying procedures. And so we didn't, you know, automate the way craftsmen by building robots that did the exact same procedures. We did that by restructuring the work in a factory setting and then applying sophisticated machinery. Then there's the question of what will the next generation of companies look like?So what we're beginning to see is AI native companies that are much leaner than incumbents.

And if those firms tend to turn out to be more productive, they may well outcompete some of the incumbents. And then we might see an impact on the labor market through that competition. Um, as well. Um, another key question is to what degree will we see these new firms emerging developing products versus processes. Right. And so for the most part, when you use a tool for efficiency and just improve productivity in the tasks that you already do, you're going to replace some workers.

But if you use it to create new products and new industries, you're going to create new jobs as well. And that's one of the key reasons why business dynamism has historically been really important to, to new job creation. So large firms, they have scale, they're more likely to use AI for process improvements. Smaller firms, young firms, they don't have scale. They're more likely to use it to develop new products and industries. And therefore their entry is going to be critical as well.

So I think what we will need going forward is better data to assess some of these trends in real time. We also need to better understand, um, how consumers are actually responding to these technologies. We tend to often focus on machine capabilities. Uh, you know, I'm quite happy to go to self-service checkout when I go to the grocery store, but when I have dinner with my wife, I don't want to sit in a giant vending machine. And so what's the difference?We need to understand in which settings humans prefer humans rather than machines as well.

And I think we lactate on that too, unfortunately.
14:35
14:35
14:36
14:36
14:37
14:37
None Portrait Speaker 4
-

I just back out a bit from the question. Sorry to. I will come back to answering it. Um, but for me, I feel as though as policy makers, it's not a question of if, it's a question of when and how. And I think we've, we're often spending too much time worrying about not worrying, overly, considering the kind of micro of this. When the macro, in my opinion, is undeniable. And I just want to reinforce that.

And I have literally no idea which jobs are going to be most augmented and which will be least augmented. And I believe anybody that tells you they have a clear view of the next 3 to 5 years is not going to be likely to be that correct. It will be different. But in my opinion, every role will be affected. But it's going to be an intentional way that we step into that future. So I believe that we can either let every role be affected, or we can make sure that every role is redesigned together and we effect the change together.

And that may sound like a slight nuance, but I'll give you an example because you mentioned tradespeople, and I completely agree with you, Carl, a lot of what you said, just to pick up on your examples of electrician, just take that further. I might try to mend things. Heaven forbid I'd fuse my whole house and blow up everybody I know could create more work for electricians, because I've actually done a worse job, could create less work, could create different work. I agree with you.

There will still be work. But the thing I believe very deeply is that there will be an electrician who is working with AI, and there is an electrician who is working without AI, and the one working with AI is going to be the one that will be more competitive, more successful, and have more wealth in the future. And I believe as policymakers, we should be tilting into a future like that for this country. We should not be spending so much time being so anxious about what the anxieties around particular sectors are.

Necessarily, we should be taking the view that this has got to be about wide participation, broad access, trying to minimise inequalities in how people are able to use it so that we don't do what we did in the last 30 years of technology, which is exaggerate inequalities to a large degree. Sorry if I slightly didn't answer your question.
14:39
14:39
14:40
14:40
None Portrait Speaker 5
-

I agree like we're in a position of radical uncertainty. And obviously we need to assess the evidence as it emerges. But I guess the challenge for policymakers, for unions other otherwise is how do we also adopt a no regrets framework that we plan for every eventuality we anticipate every eventuality, but we also think about what can we do now that we know would be good anyway?We know that we're already building a AI enabled economy in a situation where we have the lowest investment of in work, in work, investment of skills in the private sector, in the OECD, we know we have the lowest business investment in the OECD.

We have a kind of economy that's plagued by short termism, by shareholder primacy, by weak investment, by low participation of the workforce in shaping that. And I'll give you a contrast to that. I was in Sweden the other day meeting Swedish trade unionists, and one of them said something that was very striking. He said, we don't fear new technology. He said. We fear old technology because old technology makes them less productive, less competitive, less able to upgrade their abilities to participate in the economy as a whole.

But the reason why they don't fear that is because they have a stake in the in the economy as a whole, they have 88% collective bargaining coverage in Sweden, in the UK, it's like 13% in the private sector. Um, and what that means is that you have a situation where the wider system, whether it's the Social security net, the skill system, the taxation regime is actually set up to be high productivity, high innovation, but also high equity and high resilience for the workforce. So I think like part of what we need to do when we're thinking about AI, if we're thinking about it as a transformative technology, which I agree, it may be or is, but perhaps over a longer term than we're led to believe by some of the CEOs.

We need to then think about what is the institutional architecture that we need to build that reflects that. Just to briefly sum up to contrast that right now, there's 500 workers in British Gas that are about to be made redundant in call centres, because people are using customer self-serve with AI enabled chatbots. Um, and what's happened is, is that the company has introduced this, customers want it and those workers are going to be made redundant. Like that happens. It's been partially driven by AI.

Um, but there's an interesting thing which I call kind of the drudgery elimination fallacy. The workers that remain are picking up the most difficult calls, the ones where their customers are angry customers, the issues intractable. They've gone to the employer and they've said, um, well, how does this reflect then in our pay packet and our working conditions?Because we were doing more demanding work in the public sector, not going to talk about it. So that comes back to is like, how do you capture the gains of this?What's the incentive structure for the gains to be reinvested into the workforce, to be reinvested into innovation?Because British Gas over the past two years has seen a 33% increase in shareholder dividends in 2024 and 22% increase in 2025.

A lot of that will be kind of like super profits from energy prices. But how was the gains of this transformation going to be shared and reinvested in the workplace, and not just gravitate to the top?So these are some of the big questions that we need to grapple with.
14:43
14:43
14:45
14:46
14:46
14:46
None Portrait Speaker 6
-

Um. The scale of the scale of the challenge is, is real. And I think there's some very loud voices which get that part right, but perhaps not the cause or the driver, um, of it. Um, what you highlighted share about the importance of augmentation. And by that I would mean people, people, you know, people's choices and their capabilities and their skills. Um, and also redesigning work and also and business models potentially as well in a way that was directed and put certain goals first.

And people is hugely important. And, and it's, um, not a coincidence that everyone on this panel has highlighted the importance of participation and involvement, which perhaps invites us to go upstream and downstream of thinking about sharing benefits, which is hugely important, but we don't have to start there. Um, uh, Karl or other professor Frey made some very important points about other factors driving the labour market and the fact that as well, um, overall. In terms of numbers of jobs, there isn't a lot of change.

But, but um, it's, although I agree there shouldn't be an analysis paralysis. The latest research does also, um. Uh, really pull out the point that we need to refocus on work transitions and supporting work transitions. And that's happening inside jobs. That's happening inside jobs. It's happening inside firms. It's happening inside sectors, um, as tasks and skills and roles move around. Um, and I'd say there is enough evidence now, um, including evidence from the Institute for the Future of Work, um, which is an independent research and development institute, as you may know, um, that suggests that we're, we're at a point where there is, um, there are signs of a new equilibrium between that and that we mustn't assume the best results.

Um, as Baroness Fox says, it's there, the potential is there. But often those points of opportunity are also those points of risk. Um and, and that they must be understood and targeted more, more perhaps perhaps more sharply than they are now and at a bigger scale. Um, as, as we've all said, um, thanks.
14:48
14:49
14:49
None Portrait Speaker 5
-

Um, I haven't seen there be much mention or plan for that. Um, I met with, uh, drivers and reps in GMB and unite for drivers in London where obviously it's being rolled out. What I would say is that the evidence that we have so far from where driverless cars have been rolled out, um, extensively is obviously in the US and in China, but obviously the US is kind of like perhaps more equivalent to our labor market. There are job losses, but the biggest thing that we see is downward pressure on wages.

So if you look in in Seattle or other places in Phoenix, what you see is a ten around 10% decrease in the take rates of the drivers that remain driving because basically the companies are obviously going in. And it's the same model that Uber did when it came to London, is they obviously undercut the existing market because they've got capital to be able to do that and sort of like land and expand mechanism. Um, and obviously like more supply means that there's a drive down the rates of those drivers in London, there's 100,000 private hire vehicle drivers.

Obviously there was individuals, but obviously supporting families in different ways. I'm not aware that there is a substantial plan for, um, how their income is going to be affected, how they'll transition if they need to, um, or other things, for example, like the new company, the new operators, they're going to have, they're going to create new jobs. There's going to be depots for cleaning and cleaning people's sick out of it after a night out and maintenance. And the call centers and the control rooms, are those jobs going to be outsourced?Are they going to be secure?Are they going to be hostile to unions?Are they going to be kind of like on good contracts, or are they going to be further precarious position of a sector that we've seen that has been the sharp end of algorithmic management, dynamic pay, which is the sort of abuse of technology to, um.

Exploit, um, the, the pay rates that workers take, like those are the conditions of which we're arriving into. And again, that's the institutional architecture. That's not the technology. Like that is where public policy has power. And especially in markets like London, like London is a huge, um, has leverage. If any of those companies want to come play here, we should be setting terms and we should signaling what those terms are.
14:51
14:51
14:52
14:52
14:53
None Portrait Speaker 5
-

So thanks. I'll start by saying it's really contextual. Okay. Like in, you know, you can have AI deployment, for example, with automated vehicles we just discussed, like they're not listening to the employees at all. They don't need to engage them. It's not about augmentation. It's about like for like displacement in that context, right?In professional services, you're talking about gen AI. It's a kind of a different context altogether. And what we are seeing is that obviously AI is really broadly adopted, but very shallow.

It's not turning up in the statistics as the sort of famous saying goes. And there's a growing and really interesting recognition that workers have expertise, experience, and insight. There's absolutely necessary to define what is the problem that AI is meant to solve. You know, AI is often is said, like, you know, a solution in search of a problem. And boards executives also need training. Like people talk a lot about workers need to be upskilled. But like there's a lot of executives who are really drinking the Kool-Aid on this and then being like, oh, where's my productivity gains?And now there's a walk back from that because there has been a lot of spending on AI without any big productivity gains.

There's a recognition that you need people's buy in. Otherwise there's going to hide AI usage. So there's this phenomena called shadow, shadow AI or shadow white. I t uh, and you can't just have top down mandates because you otherwise have kind of this thing called adoption theater, where people are like being clocked up on how much they're using AI rather than is it actually solving a problem or solving a sort of an issue. And this goes back to the participation bit.

It's like if you bring in the workforce and you recognize the expertise, their experience in sectors where like that is valued, um, then they can help to define the problem, uh, help to define the training that is needed, design the features be brought into the system and also flag issues as it comes up. And you actually create an intelligent organization.
14:55
14:55
14:55
None Portrait Speaker 4
-

A slightly different angle declared to interest. Just stop being president, British Chambers of Commerce, and I'm on the board of multiverse. I'm going to talk about SMEs because obviously 85% of the economy, slightly different perspective to Adam's. I completely agree about context. But what we've seen at the British Chambers of Commerce is, you know, 2 or 3 years ago, I think it was 84% of those small and medium sized businesses said, I don't know what to do about AI knew it was super important in their kind of priority list, but didn't know what to actually do.

That's now gone down to 64%. I think it's now maybe around 50%, and that's partly because of programme of work the BBC have been doing, which is very specific, and it is helping people look at the problems in their business, as opposed to start with the tools. And it sounds like a very small shift. But in a small business, people are being sold, you know, whether it's copilot licenses or another. I'm not going to pick on Microsoft particularly. And they're waiting, as Adam quite rightly said, well, where's my productivity change, which just give people a tool that's not going to work.

You have to start with the problem and actually redesign the process in the business. And that can take some work and some cultural shift and some training. It's like any digital transformation that we've been through in the last couple of decades. So in my opinion, it has to be very specific, very, very detailed based around business problems, not starting with the tools and the technology and starting with actually, what are you trying to achieve?And that requires quite a lot of intention.

And especially at a time when SMEs are feeling completely up against it, as we know, pretty cross with the government, the economy, everything feeling like regulations are everywhere, feeling as though they've just had cost layered on top of them and that the projected view in the next year is not that great. Trade down, Brexit, all the stuff you guys know way better than me. So I think we just have to be cognizant of the fact that this is hard work and help and political help that can be given to companies to do that is valuable.
14:57
14:57
None Portrait Speaker 7
-

I was.
14:57
14:58
14:58
14:58
14:58
None Portrait Speaker 6
-

Um, I think we can pitch it a bit higher than that. And it's not on the basis of analysis, not paralysis that the, that, that many different disciplines, including organizational management, social sciences and economists as well, um, show that this, that, that involvement really does mediate better results in the same way that the peer review into work and well-being showed very showed very clearly that focusing on work quality and effectively good work mediated better results at different levels. So for the work of the firm and at a labor market level and, and obviously involvement in participation is a really core part of that.

In terms of involvement. It's not simple because we're talking about different levels. So we're talking about tools, jobs, um, you know, skills and pathways and then support that's in place. And there's a good case for involvement in all, all of those buckets. Um, we've got important existing infrastructures, including union involvement, which could, could perhaps in this be utilized more, but also, as Baroness, Baroness, um uh, Martha. Um. That we do need to look at it from different perspectives and perhaps look for areas of alignment and see what's happening as best practice to, um, and there are some examples of that, but they are few and far between.

So I refer to we've done eight case studies for innovate UK of responsible AI adoption in different types of firms, including SMEs and public sectors as well. Um, and they're, they, and they were basically pockets of good practice, but um, um and. And there are plenty of examples of that, of those pockets in, uh, that are, that are, that are published in those, in those case studies which were done with the Cipd. But there was, um, but there was no one case where it was systematically, uh, caught all these different touch points and all these different levels.

So I think that we need to, um, go, you know, go big and also offer guidance, not be afraid of offering guidance and support, um, and incentives to, to reward good behaviour when people are increasing levels of involvement, um, work quality and.
15:00
15:00
15:00
None Portrait Speaker 2
-

Yeah.
15:00
15:00
15:00
15:01
None Portrait Speaker 5
-

Yeah, briefly, two things. Um, so this has now been more realized that by like Microsoft have got a report that says like the employee participation is like essential for fair and effective AI adoption in their 2005 future of work report. It's not loads of case studies. I mean, I guess there's a selection bias for me at the TUC because like the problems come up, but I'm sure there are some, I guess to add to that, my annual forgive me for banging on about, um, the Nordics, but they.

Denmark has the highest adoption. In peer countries like we need to think about why it is that. And they also have like obviously a really robust economy. Obviously it's different and we can't bring everything from there over to here. Um, but for another example is that, uh, Ford Motor Company recently rehired 300 experienced engineers because their AI quality assurance that they introduced into their production line was not up to scratch. And it failed. And the executive said, um, and I'm paraphrasing, but the executive said, oh, we didn't really appreciate the skill and expertise that those engineers had before.

We kind of like basically replaced them with AI because they don't really understand what it is, the work that they do, especially in big companies where you're very remote from like the shop. Um, and so they rehired them, but then they've basically gone on to say, oh, we basically needed to rehire them so that they can train the machines and then kind of like let them go again, which is like, people will do that because you need the work, right?But like at the big picture is like, if that happens at a macro scale, that is just an extraction of, um, a wealth of experience of expertise of um, that has been gained and earned just up.

And then you get replaced by that robot. And how are you participating in that?And we've seen that particularly in like the creative industries where, um, you know, that work is being scooped up and replacing a lot of workers in that context.
15:03
None Portrait Speaker 2
-

Well.
15:03
15:03
15:04
None Portrait Speaker 4
-

I call line.
15:04
15:05
15:06
None Portrait Speaker 9
-

And just to build. Sorry to go back to the question. I'm actually coming to yours. There are so many examples that we can share offline about companies doing this well. Jaguar Land Rover pop into my mind. John Lewis, I know them through being on the board of multiverses. I've declared the interest. They were making a bunch of people redundant. The people that they kept in the business, they designed how their jobs were going to look using AI and gave them AI training to do that.

And that was hugely successful. Project capita, when they were collapsing. New boss comes in. The first thing he says is anyone in the company can redesign any bit of the company using AI?Tell me directly as the CEO, and had a huge cascade program. So there are lots and lots of examples and I'd be happy to share them offline. Um, comfortable management point, of course, but I completely agree with Adam. This is just about good management, full stop and good leadership.

I'm not sure there's like good AI leadership. I think it's about recognizing that you should always engage people in any change. It's a cultural shift as much as a technical abilities shift. Um, and when I think back to the digital transformations I've been personally involved in, that is always the bit that is harder because it takes a longer time, you're going to inevitably have to shift and move roles, people, and there will be change and that can be uncomfortable. So I think, uh, I'm not sure I see a specific AI piece of this.

I think it is just about good management and good leadership. But I also totally recognize that we have a challenge in helping leaders understand the art of the possible right now. And, you know, that's true on a 5000 board, in my opinion, as it is true in a small, medium sized business, A, B, C, C member. And so there is work to do, I think, in helping people feel confident, helping people feel like they know the the art of the possible. And I think that that the two are related.

But that's.
15:08
15:08
None Portrait Speaker 9
-

Yes.
15:09
15:09
15:09
15:09
15:10
15:10
None Portrait Speaker 5
-

Around algorithmic management in particular, which I think is distinct to this kind of broader question of task automation. So if you think about algorithmic management as essentially digitalisation or automation of the sort of human resources employment relationship, as, you know, um, because you participated on the TUC AI bill, um, um, uh. Working group. Uh, I guess the short answer is no, we don't think that, um, we, the TUC drafted a model piece of legislation that set out, um, how to, uh, undergird and this has come back to a point I was going to make earlier.

Good culture with the right incentive structure or in short, the right carrot and sticks if you just rely on good culture, the good guys you want to go take the high road, get undercut by the ones who don't like, you know, the, the logistics company that doesn't want to like, rinse their employees for everything, get undercut by the ones who don't. Um, and they can use technology. We need to set a floor, which is why we've called for legislation on this point. Um, there is the, um, in the government's make work pay commitments.

There was the workplace monitoring technology. Um, component. Actually just been released now for consultation today. So we'll be feeding into that. Um, I think what we're saying is that that process of algorithmic management breaks down the human relations component that we've just discussed is really important for like effective, dignified, transparent, empowering workplaces. If people are kind of like managed predominantly by algorithm. Um, the other bit that I would mention is the interface increasingly between algorithmic management and task automation, which we're predominantly talking about today.

So, um, we know that, for example, workers are being asked to wear forward facing cameras when they're working in logistics or in factories, because to collect training data for robotics, automation, um, obviously, you know, drivers data has been used for automation of vehicles. And so like it's interfacing increasingly in a way in which algorithmic management is for algorithmic algorithmic data collection in order to produce the training data for task automation. And this is a new frontier that we're interested in exploring because with large language models, obviously, the internet has been the training data basis for robotics and other things.

You need to collect other bits of data. The question always come back to is like, how do people participate in and gain from that transition?In the workplace or are supported through disruption in society. Um, but to come to your answer no, we think there's a lot more work to be done there to, to raise the floor of good management practices, essentially.
15:12
None Portrait Speaker 9
-

Yeah. I mean, I, again, agreeing with my friend Adam here, I think that it is still somewhat surprising to me that in the last 30 years of my dabbling around on the edges of technology, there has not been substantive regulation, really, of a in a technological sense of any of the new kind of waves of technology. And I mean that very broadly, um, apart arguably from the Online Safety Act. So I do think we need to reckon with what is that going to look like in the next 5 to 10 years.

And I don't know what that answer is, but I would completely support two things. Firstly, I think we should really make sure that the regulations we already have, we're deploying well across the things that are happening. I think there are a lot of things that we could probably make sure we use more effectively that we have already in existence, rather than creating more. I think we just have to always have that backdrop of that small business from the P.C.C. On my shoulder saying, no, don't make us do more stuff.

But you know, there shouldn't be a, there should not be a, um, too much of a challenge there. Everything should be additive in my opinion, and helpful and make sure that there is more productivity and more growth in the economy. So we should make sure what we have already works better. And, um, there's no question that I think that something about the algorithmic opaqueness that is unhelpful generally. And I think just as a very like entry level 101 to some of the laddering that Adam's been talking about, I think that is something that feels like very low hanging fruit, just that everything is transparent, transparent, as opposed to opaque how you're using it, where the data is flowing, what's coming from you, what the, you know, when I'm sending in my CV, what's that transaction as a human being?And how do I understand that?
15:14
15:14
15:14
None Portrait Speaker 3
-

Yeah, exactly.
15:14
15:15
15:16
15:16
15:16
None Portrait Speaker 2
-

Um.
15:16
15:17
15:17
15:17
None Portrait Speaker 11
-

I'll go.
15:17
15:18
15:18
None Portrait Speaker 5
-

Uh, I'm not sure I could authoritatively answer on that question. I'll say something briefly about the other other component parts, which is outside of the gig economy, which is obviously different places, basically like when workers as individuals or as a collective bargain for pay, their bargaining time and skill, essentially. Right?Or some variation of that. We know that, um, AI can uplift people's skills in certain contexts. Absolutely. Because people can access new information they can access, as Carl mentioned, kind of like new content, including kind of tearing down barriers to information.

That's kind of like usually been the preserve of sort of like cloistered professions, but we also know it can have a de-skilling component. So like anthropic, the makers of Claude did a study where they found that they put two sets of computer programmers. They gave them a task, one used AI, the other didn't. They found that the ones that used AI had a 17%. This is their metric decline in skill mastery, and that's the process of cognitive offloading, right?So you've you've basically given more of the value of what you've produced, more of the value of what's been produced into the machine, which means that therefore you've been deskilled, which as an individual or as workers, as a group, you have less collective bargaining power when you go into the workplace.

And that's been a process. There's obviously been happening for quite a while. And so the question for us, which brings us to the skills question and other things, is like, how do we make sure that AI is deployed in a way that actually uplifts people's skills, rather than sort of like atomizing or degrading or cognitive offloading their skills and therefore having the impact of diminishing their individual and collective bargaining power. Final thing, because I've been talking for a while, one of the things we're seeing is kind of like wage stratification or wage polarization.

So for example, computer programmers, which is really the only place where AI adoption has been really significant in like kind of systematized, uh, experienced programmers are seeing broadly significant increases in pay because they're able to boost their productivity by using coding assistants and, um.
15:20
15:20
15:20
15:21
15:22
15:23
None Portrait Speaker 13
-

Um. I.
15:23
None Portrait Speaker 11
-

To start slightly differently. And that I think that is a danger in where we are right now with the journey of AI and us culturally in Britain that we have ignored some of the things that might be important and kind of overplayed some of the things that might not. And I'm going to pick a specific example, which may be hopefully wraps up some of the things we've been talking about. So if you look at young people's fear of AI, it's obviously disproportionate to other bits of the population for understandable reasons.

They're sending 45 CVS that they may have made themselves using AI, but they're all being algorithmically chucked out at them. And there's a disconnect between their view of the world and what they're seeing and how AI is affecting them personally, and what we all probably believe is going to happen and be productive and helpful in the economy of the future. And I think, obviously, that is a very, very profound challenge, because if we have a whole group of young people, both the Melbourne cohort, but then another group of young people who may well be skilled right now and may be trying to find work that are unable to do so, then that's a real social cohesion point.

So I think I approach it from a slightly different perspective. I think I'd say, okay, bit to Adam's point, what are the no regrets things?Who are the cohorts of people right now?What are the messages that we as government need to land about this stuff?Because we need to make sure that the UK is a place that is not only building the AI sector, which I think we've done incredibly effectively. I mean, we've done an amazing job at becoming the third biggest place for investment into the AI sector after Silicon Valley and China.

It's remarkable. London and the UK should be really proud of what we've been building. That's a lot to do with DeepMind, diaspora and all the things you've I'm sure thought about a lot. But I think we need to, in parallel to that, to think about how are we going to message what it is we're going to do for the broader economy and specific chunks of it?So that's one piece of it, which may sound kind of, um, slightly. Flaky, but it's not meant to.

It's, it's, it's actually quite deep, difficult work of how government joins up all the different bits of kind of business support, all the different growth messages to always continually landing that this is the direction of travel, this is the way we're going. These are the things when we make an intervention, we're going to do it because of an AI first world. And I'm not going to go back and re legislate stuff. But the employment law that just went through that was not designed for an AI first world.

And I just think we need to be really cognizant of what we're doing with intention in this building all the time about what are we really building for 2030?Or are we building for 2025 or even worse, 2020?Or maybe we're not really even building for it at all. So I think that's the first thing we need to upskill as many of the people in this building as we can to what we all believe is a view of what this future could be. Pick off some cohorts that are in, I would argue, um, more danger right now.

And I think entry level jobs is one of those places not just to do with AI. It's clearly to do with the macro economy as well. But there's a kind of horrible mix going on right now with a whole bunch of things together. And I think the fundamental underpinning of all of this stuff should be all the programs that government do should just be questioning about, are they designed for an AI first world?Are we really helping people?When people go into a job centre, are they sitting opposite a desk at someone tapping away on a computer, telling them something, or are they sitting behind the desk with the person co-designing their CV able to use the tools themselves, able to see what different roles, what those transitions and pathways that Anna talks about so well might be.

And I. So I'm not avoiding your question, but I just think it's not one thing. It's about every single signal that government can send. And having a really joined up and collective.
15:26
15:27
15:27
15:29
15:29
15:30
15:31
15:31
15:31
15:32
15:32
15:33
15:33
15:34
None Portrait Speaker 4
-

Please continue.
15:34
15:34
None Portrait Speaker 15
-

So I mean, I think, um, there's a lot of what, um, Julia said that I would agree with and, you know, I was I'm a computer scientist. I was a computing teacher. Um, before I moved into sort of trust leadership and, you know, for sure the current national curriculum is out of date. And there's a real risk that the next one does the same. If we're if we're not careful. But I would really caution against the urge to over prescribe what the new national curriculum is.

And I think it's one of our biggest concerns is that the programmes of study become too Overprescribed. Actually, we have got schools doing a really brilliant job of teaching digital literacy and AI that are sort of tapped into what's going on in their local communities. That's based on the experience of children and what families are telling them about their children's experience of technology. Um, and that is guided and supported actually by, by the trust and by networks that people are part of your organisation, charities that operate to provide pupils opportunities to code, um, from my perspective, we don't offer actually a great deal of qualifications around computing, computer science.

Our focus is on building powerful knowledge across the whole curriculum, developing pupils broader skills, critical literacy, and their ability to think increasingly through systems thinking, their ability to speak well because we recognise that actually those skills are the sort of primary skills that employers are going to want ahead of perhaps the ability to code. Um, when you can vibe code or not. I don't want to undermine the sense that we want to understand the underpinning technology. And young people need to know how this technology works, but they don't all need to be able to, um, get a GCSE or an A level or pursuit ahead of perhaps the ability to code.

Um, when you can vibe code or not. I don't want to undermine the sense that we want to understand the underpinning technology. And young people need to know how this technology works, but they don't all need to be able to, um, get a GCSE or an A level or pursue a computer science or AI degree. So I'm more relaxed about the time it's going to take for PCT ahead of perhaps the ability to code. Um, when you can vibe code or not.

I don't want to undermine the sense that we want to understand the underpinning technology. And young people need to know how this technology works, but they don't all need to be able to, um, get a GCSE or an A level or pursue a computer science or AI degree. So I'm more relaxed about the time it's going to take for qualification reform. I think, on national curriculum. I think that my I would encourage us to be as lean as we think is sensible to enable schools and trusts to use their local labor market information and broader context of the children and communities, and what's happening now.
15:36
15:37
None Portrait Speaker 16
-

Yeah, I mean, I, I agree with what both Lauren and, and Julie have said. I mean, I think the key to this is that as was articulated by Julia, that you've got, you know, a system which is clearly got the pressure of quality and consistency and recognition and so on. And I mean, I'm not an expert in the school system, but you see the same in the apprenticeship system and in the even the degree system to agree. And then you've got this kind of alternative dynamic that of course, is, is, is very strongly visible in AI of rapid development of technological evolution, which is way on a different clock speeds, just use a computing terminology than, you know, the education system.

And so I think we have to try and find ways where and Lauren put it very well, because I had an example I was up seeing last week, seeing Amy, you know, the big construction company. And I said to them, I said, are you guys finding AI coming through in The Apprentice, creating a lot of apprentices, T levels, everything. Are you seeing it coming through?And he said it varies depending on who we're getting to do the training. So Lauren's point is that, you know that in many cases, trusts and others are doing this kind of because they're good trust and they think about the bigger picture and so on.

Um, and I guess the balance we have to get to is to how you bring everyone up to that level and, you know, exchange best practice, get people lean and dynamic, get people and particularly educators and maybe some of the, some of the curriculum structure better so that they've got the space to actually do this. But it is it's a difficult balance that that's struggled to because in previous iterations, of course, the, the cycle was probably acceptable generally as digital technologies come along more strongly, it's become more and more difficult.

So, um, yeah, I think it's an issue we see everywhere, but it's definitely one that we need to find our way through. And you know, we'll talk more about it. I'm sure as we, as we talk about, you know, adult skills. Thanks.
15:38
15:38
None Portrait Speaker 17
-

To add?Nothing to add.
15:38
15:39
15:40
15:41
15:41
None Portrait Speaker 4
-

So thank you.
15:41
15:41
None Portrait Speaker 14
-

Um, so I think, you know, the, there is definitely the, the, the need for young people to have those sort of traditionally termed softer, softer skills, um, more employability skills alongside the technical skills. I think, I think there's a sometimes perhaps a misalignment in expectation from employers about what an entry level job should include and actually perhaps in some way is more senior than perhaps traditionally was with perhaps in some places, AI, um, taking on some of those more tasks associated with a sort of fresh graduate role, for example.

Um, one interesting point I think to make is that, you know, we know that undergraduates really benefit in employability skills. So when they've got a placement or internship as part of their course, they are more employable. But actually perhaps not, not as well known is that when employers, um, take on, uh. Placement students, they are then more likely to take on young people, even young people that they haven't met. So their attitudes, their culture, their hiring practices, shifts. Um, and so there's something there in that sort of, it needs to work both ways and we need to support and encourage and incentivise for it to work both ways so that employers are actively.

Seeking, um, because these young people have got a lot to offer. Um, and they are, they, you know, I negate the term digital native because I, you know, no. Um, but um, there's certainly attitudinally ready and up for, um, developing their role with AI and being able to add value to the business. So I think thinking about changing attitudes of hiring employers as well is important in all of this.
15:42
None Portrait Speaker 4
-

Thanks very much.
15:42
15:43
None Portrait Speaker 16
-

So the system is evolving. Um, there's no doubt about it that, um, but you know, if you were asking whether it's completely in place now, then the answer to take the chair's view is no, it's not appropriate to do that today. It's evolving because we're providing things like short course, um, capability, for example, you know, a so-called apprenticeship units. Um, but if you look at the work we've also done with the technology industry and the Department for science, Innovation and Technology, um, in that.

We're working with vendors to take their capabilities and provide them, um, to young people with a framework around it to, to people generally, frankly, with young, um, with the capabilities around it to say this framework ensures a certain amount and we can give it a certain badging to give it some credibility. So I think there's a lot of work going on both in there. There's AI unit, there's AI apprenticeships that have been created on fast track. There's AI in a number of the standards already, but of course, a lot of the standards are 700 odd standards.

So there's a huge number of them and they have a level of stability, as was described before in the, um, in the curriculum context that we need to evolve. But of course, you don't want to throw the baby out with the bathwater and crash apprenticeships, you know, um, like happened before with the, with the levy. So, so I think it's got a long way to go, to be quite honest. And part of it is, I think the discussion you were having before about the whole system being more joined up, because there's quite a lot of fragmentation even in the AI space today inside government.

And I think that needs to be pulled back together. If we're talking AI specifically. And then obviously bringing in digital skills more broadly.
15:44
15:44
15:45
None Portrait Speaker 13
-

Okay.
15:45
15:46
None Portrait Speaker 16
-

It varies, and I would slightly dispute, I think, yes, I agree, there's a lot of noise from employers in there, but I think Julia mentioned it earlier. The employers who actually do take either young people on or apprentices are incredibly convinced of the value of apprentice. I mean, Amy, again, just as an example, but XL motors um Coventry I went to see the week before. I mean there's just lots of them. They completely buy it. But I do also agree there's a bit of a hurdle to get over.

However, if they then said is that apprenticeship in, I don't know, plumbing?Is that actually now preparing me for AI?And I think the short answer is probably no. There are short course. Um AI upskilling modules available, which are 30 hour modules which can be funded by the levy. Um, so three of them that are being created. Um, so that's, you know, a start of that, but also if you look at things like the AI boost programme, which is the commitment from government to, to train 10 million people that's available free to use, um, curriculum or capability that can be used to supplement things.

And, you know, I was in number ten this morning with a bunch of people talking about AI, and it was quite obvious that even in the short that we've had two meetings, three meetings in that the things that people wanted to teach three meetings ago are different than the things they wanted to teach now. So three meetings ago, they were talking about, you know, um, prompt engineering and so on. And they've moved on from that. Now it's about agent management and contextual or context management.

So, so trying to keep up with the technological advances here in some sort of government context, frankly, is impossible. Um, but I do think government needs to put the framework in place to allow people to do this in a more dynamic way.
15:48
None Portrait Speaker 17
-

Um.
15:48
15:48
None Portrait Speaker 17
-

Yeah.
15:48
None Portrait Speaker 14
-

Yeah, I do on that point actually. So, um, you know, trying to keep up this just in time training is going to become, you know, probably the model, uh, but trying to keep up is, is, as Phil says, probably fruitless. Um, so actually, um, in BCS, we run an annual tech priority survey amongst our members. Um, we ran it in 2025 and the committee has already got that, that research available to them, but we've got a live 2026 one running at the moment. So I've got hot off the press information and on unsurprising, I think, to know that AI skills is the top in demand cited gap.

Um, and, but actually the strongest signal we get from respondents is that anyone using AI in work should have completed a recognized formal ethical certification. So why we won't be able to teach necessarily every bit of how this technology works and how you're going to do it. In that sense, if you've got a workforce that has really good groundings in the ethics, they understand responsible use, they understand the risk that they can introduce, they understand the opportunities and the benefits, but they are, um, critical thinkers in their use of AI.

Then we stand a chance. And I think that's what needs to underpin the curriculum. And that's what needs to underpin skills and training, not just at school, college, university, but right the way through. And we need to get our heads around that cultural change. This isn't just lifelong learning as a as a as a phrase. It is literally lifelong learning, underpinned by that central sort of ethical, um, responsible use for everybody. And one final point, you know, I can't stress enough.

And the other panel were saying it too, that contextualizing the learning for the sector, the region, the role you're in is where you make it sticky. And that's why at BCS, we work lots with teachers. We've got an AI confidence programme. It's all it's all built on what they experience in the school and in the college, because it's different from generic IT training, which kind of most of us, you know, maybe we don't have an affinity with it. We're kind of we either are tech or we're not tech, and we don't want to do tech training.

But if it's about my real job, the real children in front of me every day in the classroom, and it's about how I harness that technology for better educational outcomes, then I'm going to invest my time in it.
15:50
15:51
None Portrait Speaker 17
-

Yeah. Thanks for having me along to do that. So just, just for context, so I work for an organization called Nest Insight, which is a public benefit research innovation centre. It's part of Nest Corporation, but with a sort of distinct public benefit remit focused on financial security for low and moderate income households. So the, the issue that we've been working on addressing is the very low levels of cash savings. A lot of people, particularly low and moderate income households, have, um 1 in 10 UK adults has no savings, no cash savings at all.

A further 20% have less than £1,000. And the solutions we've been looking at, as the chair says, have been around payroll deducted workplace emergency savings tools. So these are essentially tools that employers can offer to their workers that enable them to save through payroll. Not not unlike workplace pension savings. So the money goes out before it hits your pay packet and builds up kind of automatically in the background. Um, I should say. That, you know, clearly the immediate focus for us has been on cash savings for emergencies.

How do you help people have a bit of resilience against sort of short term unexpected expenses. But we have seen that there are a number of wider potential applications of this work. So that includes things like supporting longer term wealth building and retirement savings behaviours. It includes supporting agendas like child poverty, which has a financial resilience component to it, social mobility, which has a financial resilience component to it, and also actually violence against women and girls, where this this kind of model can help people build flee funds.

So we've sort of got into it through an emergency savings, um, sort of lens. But it does have this kind of broader suite of potential applications, I think, in terms of how that relates to the work of this committee and this inquiry, I guess. So we have already seen some ways in which these kinds of accounts can help people mitigate some of the labour market impacts of AI and AI. AI like challenges that can include things like support with the costs of relocating. If you need to do that to find a new job and support to bridge gaps.

If you have a week without earnings because you're changing jobs and actually things like meeting upfront costs of new childcare settings or new rental, if you need to move geographically for, um, to move jobs. So having this kind of buffer fund can help labour market mobility in a number of ways. We haven't looked because it wasn't kind of how we got into it. And we're still at relatively low levels of scale in these models. We haven't looked specifically at questions around retraining.

Um, but what I would say is I think there is some evidence that these, these kinds of accounts could, could support that model. So, so partly that's just to reflect how they're designed. So these accounts are flexible. They're accessible to people when they need them for whatever purpose they might want them. There's no cap on the amount of money that can go into the accounts. Um, some of the employers that we've worked with have expressed the kind of idea that actually helping people with their financial resilience can help them be more open to retraining, open to, um, seeking to progress through the firm because it takes away the kind of anxiety and concern of that, those low levels of financial resilience.

And we've also just actually seen some evidence in the US that, um, uh, helping people with financial resilience can reduce drop out rates from things like apprenticeship programs. So actually retraining and then encountering a financial emergency can drive drop out. And so if you have some kind of buffer fund to go back to, you actually see higher persistency rates within the retraining programs. And so in apprenticeship programs in the US, that's helped as well. So, um, you know, I would say that the, the models that we're trialling, um, they're still relatively low scale in adoption terms, only about 7% of UK employers offers a mechanism like this to its workers at the moment.

And they tend to build quite slowly because they're targeted on low and moderate income households. So contribution rates into these accounts are not always that high. But I think certainly as we work to scale adoption of them and people's savings balances build, you can certainly see a wider set of potential applications for how they could be helpful.
15:55
15:55
15:55
15:56
15:57
15:57
15:57
15:58
15:58
15:58
None Portrait Speaker 4
-

You just can't.
15:58
None Portrait Speaker 14
-

Get some of the.
15:58
None Portrait Speaker 4
-

Retraining they need.
15:58
15:59
15:59
16:00
16:00
16:01
16:02
16:03
16:03
None Portrait Speaker 4
-

Yeah. Perfect. Julia.
16:03
16:04
16:05
None Portrait Speaker 4
-

Lauren.
16:05
16:06
None Portrait Speaker 14
-

Go for it. Um, so I think firstly, don't overestimate the baseline. So future now one of our partner organisations, um, notes that over 60% of the UK labour workforce can't complete the 20 essential digital tasks for, for work, there's over 3 million people in the UK that can't complete any of them. And there's no region with more than half their workforce, um, able to complete all 20. And these are tasks like, you know, online banking and changing your password and setting up an email account.

It's basic essential skills. Retail sector being the weakest. And it cuts across all age ranges. Um, so it's not an older person's issue. Half of 18 to 24 year olds can't do these tasks either. So, you know, and those people often find themselves in SME. They may be lower paid or insecure workers. Um, so it's important not to overestimate, you know, we talk about AI a lot. So we think about five coding and advanced tech skills, but actually we've still got a problem with basic skills for a lot of people.

And whether that's attitudinal capability, interest, um, or moreover, perhaps thinking that everyone else knows more than they do, and they just need someone to get on the side of the desk with them and do it with them. Um, so I think for SMEs, bite sized, um, embedded, low barrier to entry, don't assume knowledge, um, using CPD models that go through trusted institutions like I described earlier, um, are going to be really important, um, to keep pace with the speed of technological change. We've got to go with the, the SME and their, their business and their business priorities.
16:08
16:09
16:11
16:11
16:11
None Portrait Speaker 18
-

Very happily. But I appreciate the generous start. Um, I think. At a very high level, I think there are 2 or 3 central choices that countries can make when they think about a vision of growth and its relation to artificial intelligence. One is the question of the extent to which you want to focus on being in the frontier of invention rather than, uh, the sort of long tail of adoption. Uh, and then the second choice, which is really critical, is, again, the extent to which the focus of adoption is primarily focused entirely on, uh, particular parts of industry without kind of as much of a focus on its spread.

So the absolute scale of adoption rather than the spread and qualitative nature of it. And if I take those two dimensions, uh, the broad read I have is that the United States, for example, is extremely focused on the first question on the frontier and a bit less on adoption countries like India, from my anecdotal experience, have been very focused on the development of small, application focused models that can be put in the hands of millions and billions of people, and so much more. Focus on, uh, the, the sort of long tail of diffusion rather than the frontier of invention, uh, and whether UK sits in my mind on that question is effectively taking a very targeted approach to saying there are some very selective areas of frontier invention that are really critical to both productivity and national security.

And we don't want to just give up on that. And so we're focused in those areas and we can talk about what those are in a second. And at the same time, the bulk of our focus has been on driving adoption and diffusion. On the second question, which is, when you think about diffusion, how do you think about the scale and scope of it?Effectively, in Britain, we have taken an approach where, of course, we want to move the needle in aggregate on productivity, but we want to do so by taking the country with us together.

Because my belief and the government's belief is that you only build a durable consensus behind technology in this wave, unlike the last, if the upside of that is spread across places and people as well. And that really underpins a huge amount of the approach we've taken both to infrastructure and wider AI adoption.
16:13
16:13
16:14
16:14
16:14
16:15
16:15
None Portrait Speaker 19
-

Going to take a slightly longer approach to answering this. Let me first back out and say, what is the point of sovereignty?What are the outcomes that determine whether you're sovereign or not?And to me, there are two core outcomes that I'm really interested in that have defined our approach to British sovereignty on AI. The first outcome is you need to be sovereign in having access to critical inputs that are core to your operation as a country. If you can't have Nvidia chips, you actually can't do inference workloads in AI.

And as a result, you're not truly sovereign. We're not going to overnight or even over a multiyear period build Nvidia chips in this country. So one definition of outcome on sovereignty is can you access core inputs?A second definition complementary to it is can you shape this in the light of your values as well?And I start with those outcomes because it then defines how you get there as a country of the scale and the strengths that Britain has. To me, there are three parts that determine whether you can achieve those outcomes.

And I call this sort of broadly a ladder of strategic leverage that we have to build. The first part is for core inputs that you can't develop in this country. You just got to get enough of them today. So Nvidia chips, I mentioned, one of the first things I did in this job was to go and negotiate for 100,000 GPUs, the largest deployment of GPUs in Europe to be in the UK, because whatever else you want to build here, you want to build national security deployments, core built in Britain.

You can only do that on those chips. And so to be able to secure enough is the first step of the ladder. But without it, you can't kind of go anywhere else. The second step is don't just secure it from one supplier, but try and diversify so that the core insight of Labour Party politics, which is that bargaining power and who has power matters as much as what you do with it, is deployed in our strategy here. And so I went to see Grok and Cerebras and Sambanova got all of those other chip companies, including British chip companies, to accelerate their engagement with Britain so that over time, we're diversifying where we're securing supply from so that we are building more leverage for this country.

And then the third step is, of course, in targeted areas to build full fat British. And so we have a very specific set of areas, inference chip design, which is where the future of chips is going. The series of Non-language Frontier intelligence models that I described earlier, and then some very core areas of application, life sciences, cyber security, the services areas and national security alongside trust and safety. And so that is effectively the package we're building all together in achieving the two outcomes I described.
16:18
16:18
16:19
16:19
None Portrait Speaker 19
-

Conversation we've just had, I Grossman's a really critical part of Britain building his sovereignty. And I say that in light of the ladder I described, um, if you don't have the requisite level of compute in this country, we're not trying to sort of blow the lights out on this. As you all know, the United States has 75% of data center capacity globally. Britain barely features on the table. We haven't built very much compute at all. But if you don't build the compute in this country, you can't build homegrown companies that are in the NHS.

You can't build homegrown companies at the heart of national security. You can't build homegrown countries driving industrial and operational technology productivity that I know you've heard of, heard from companies on as well. And so to me, it is the foundational layer, which is the first ladder. In some cases, you have to get inputs from somewhere. The reason most of the value in compute data centers goes to the United States and actually ultimately in large part to Taiwan and Korea too, is because those chips that are the heart of it economically, 7,075% of the value accrues to chips in data centers are accruing to companies elsewhere.

And so, of course, in the short term, we have to do that because that is the foundation on which you build sovereignty in other parts of the stack. But alongside that, as you all know, we launched this hardware plan because what I don't want to happen is in 3 to 5 years time, we're still sitting here and we're still building data centers just with foreign chips. I think the puck is going towards more and more inference rather than training. And there Britain can make those chips.

In fact, I think it's our best bet. At Reindustrializing using the power of private investment to be able to build chips in South Wales, which has strengths in Edinburgh and Scotland, which has strengths and ultimately in parts of the North, and Yorkshire, which has strengths in chips as well.
16:21
16:21
16:21
16:22
None Portrait Speaker 19
-

Reasons, chair, that I try not to comment on anonymous briefings is that because I find them mostly inaccurate and I find them inaccurate on two counts here. One, it is not the case if the AI growth Zone program is the focus of this particular question, then most of that has been driven by public financing. In fact, almost none of it has been public financing. It has effectively been the public sector supporting through proper supply side support, i.e. permitting and planning, acceleration and rationalization, very significant amounts in this case, 28 billion plus of private investment alone.

And so there's been very little public investment that has driven that program. The second thing to say is that this doesn't run counter to a spirit of devolution. In fact, it only affirms it. The nature of these programs was that these were plans, bids developed by local mayors alongside local authorities, alongside the local university, alongside local public services and yes, alongside in some cases, both local private private companies in Scotland. For example, data via a brilliant Scottish provider alongside international capability that other firms brought as well.

And to me, that approach a very cluster specific approach in the northeast, driven by the strength of financial services in that area, anchored by Sage working with the mayors and local authorities in Scotland, driven really very much by the super semiconductor industry, anchoring the rationality for why data centers are needed in South Wales. Again, the compound semiconductor cluster and work driven by the Cardiff Capital Region, primarily in North Wales, the aspiration of clean energy and small modular reactors being at the heart of why data centers are being built there.

In Culham, the home of our nuclear fusion Research Agency, a public sector organization. At the heart of this is a very cluster specific, devolved approach that has been the driving force behind the AI.
16:23
16:23
16:24
None Portrait Speaker 19
-

So this is the central question, right?I read the Pope say that technology ought to remain human. And my sort of personal motto is to consistently in the realm of technology, focus on machines that serve us rather than the other way around. And if that's your primary lens to think about this question, there are lots of ways in which we have to be very clear in our choices to support that vision. One choice is who makes this technology. And one of the reasons that I've been really passionate with the Secretary of State in making sure that in this wave, unlike the last, the opportunity of fundamentally making this technology is in the hands of millions and millions of people in this country, women, people right across this country, people from ethnic minority backgrounds, kids growing up in places where their prior generations might not have had opportunity is because of that.

So what are we doing?We are spreading opportunity. As you'll have seen, not least through the major charter that we've now signed, a series of largest academic organizations. Women will have appropriate rights in the workplace when they're doing research jobs. That's a huge landmark advance for opportunity for women. What we've done on open source is another part of it. One of the things I'm really proud of is that Britain has led the past wave of open source software, which means that a kid growing up in, say, Barry in my constituency, is able to develop software on a Raspberry Pi at almost no cost, and is able to then play a major role in shaping the future of technology.

I want that to happen in AI as well.
16:25
16:26
None Portrait Speaker 19
-

Thank you. So in the last wave of technology, software was made in one of two ways closed source, which effectively meant that large companies primarily developed software and retained all the IP. And it was a shut ecosystem. So to be able to participate in it and build apps on top, build new applications with that software, you had to purchase the IP and the license. And so there was no way that a child sitting in South Wales without the economic potential could have done so.

There was no way, by the way, that someone sitting in, say, rural India or rural Africa could have done so. What happened alongside it was there was an open source movement. The web, for example, was open source. And what these did was these were protocols which said we are not focused on shutting the IP. We're focused on building an ecosystem, letting a thousand flowers bloom, people baking on top of this without having to pay for the license. And that was phenomenal because what it did, and Britain was actually the leading frontier of this in terms of just the sheer scale of open source talent, was that a child can now, for a few pounds, effectively build on a small Raspberry Pi.

Aspects of software and learn what that looks like so that they could then build applications that were remarkably useful. And this meant a huge spread and diffusion of technology in the last wave. If we can capture some of that opportunity in this wave, I think we will have done a service not just to adoption, but to openness of adoption by people who otherwise wouldn't be part of it.
16:27
16:27
16:28
16:29
16:30
16:30
16:32
16:32
16:33
None Portrait Speaker 21
-

And that is the opportunity, I think, in front of us in terms of the world of work, continues to change as technology continues to change and people's experience of that in the workplace. Um, so that we can, if used well, um, for businesses and their employees alike, that it absolutely leads to better productivity, the growth in wages growing across our economy, work productivity and flexibility at work. There's a real positive opportunity for us to grasp if done well and if deployed well. Um, and that's absolutely what's at the forefront of my mind is, you know, when we're approaching this topic, how we can make sure it does absolutely enhance, enhance workers and growing wages is obviously a real key priority for us as a government.

As we grow our economy, improve living standards all across the country, work is absolutely essential to that. And of course, as tech changes and impacts their experience of work, that that is the future that we see and and Blair have touched. On that trust element and obviously the adoption piece in terms of businesses, and I think trust is really vital to uptake of how tech is used in the workplace. And with that. And if we bring the workforce, if employers bring the workforce with them, and there's a real win win here that we can actually use it responsibly, help grow our economy, use it productively, and that businesses can really reap the benefits of it.

For example, using AI tools to provide better service and increase productivity for that business alone. But absolutely, the question of who is benefiting from it is, is right at the forefront. And a key question for for us in DVT and in my work regarding the wider employment rights picture and space that that successful adoption has to have, bring workers with us. And absolutely, they have to benefit to.
16:34
16:35
16:36
16:36
16:36
16:37
16:37
16:37
16:38
None Portrait Speaker 13
-

That bit, creating.
16:38
None Portrait Speaker 21
-

More rewarding jobs, paid good quality work, which is all part of our wider make work pay agenda and feeds absolutely into this, um, into this question to supporting people, going through jobs, transitions and supporting young people into work. Technology for me is absolutely part of that picture of how we look at the labour market in the 21st century, and how work continues and will always continue to change, and how we keep pace in terms of the rights. Um, that's, that's in my brief. So I suppose on the, how, um, how we're looking at this and exploring it and Kanishka will touch on lots of work in detail around the economics Institute and that cross-government working that's taking place.

And DBT are absolutely feeding into that in terms of the AI and future of work unit, to make sure that across this whole space, we are prioritising that secure, good quality work throughout. But we're also in terms of the work of voice piece, um, absolutely central to all of this is ensuring that our digital transition and adoption is working with employers, with trade unions and, and workers, and particularly on the workplace monitoring side of things, which is one experience of AI and how it's used in the workplace, how we continue that dialogue with businesses and workers.

So we've launched a conversation and a consultation about that to build that evidence base of how it is being used, that we can ensure that running thread that I've talked to already, that workers are absolutely benefiting from these gains, that we do have that smooth transition, that we can accelerate adoption, but use that best practice throughout. So this consultation that we've launched will be really key to building that evidence base and understanding what is actually happening in workplaces.
16:39
16:39
16:39
16:39
16:41
16:41
None Portrait Speaker 20
-

What's that?Sorry.
16:41
16:41
16:41
16:42
16:43
None Portrait Speaker 13
-

Um.
16:43
16:44
16:44
None Portrait Speaker 19
-

Models we've. I think to my knowledge, we've evaluated pretty much. I don't know when the first model generation was. But subsequent recent models. We have effectively evaluated all the core frontier models across the companies that I mentioned. And so then the question you have to ask yourself is a system that is putting the UK in such a distinct position, which is that regardless of the fact that none of these country companies are in this country from an incorporation or training point of view, are doing this because we have used state capacity as the driver of the outcome rather than necessarily regulatory specification.

It's delivering. The outcome we're interested in is delivering it uniquely of any country in the world. And so what is the rationale for switching from that mechanism to an alternative mechanism?And so I'm not saying I sort of have the answer on this, but the big thing I would say is that the reason we've been able to secure the very distinctive outcome to protect the British public and secure British interests is because of the capacity and capability we've built, working with the companies. And so, you know, of course, I'm not at all ruling out that putting that on a footing of statute at some point might be a helpful improvement or not.

But the reality of it is the outcome is what matters. We're obtaining that outcome. And anything that changes our mechanism of obtaining that outcome, I think, ought to be evaluated in light of that.
16:45
16:47
16:48
16:49
None Portrait Speaker 13
-

Um.
16:49
None Portrait Speaker 19
-

So one question is just, uh. The scenario described is obviously one of the scenarios. And so I want to kind of caveat it by saying that there are a range of different scenarios. I think the scenario you're describing in particular of existential risk being obtained is a pretty low probability scenario. So with all the caveats, but I want to take it very seriously because the impact is so high that even if the probability is low, you want to take it seriously.

And there are a couple of different things that I think are appropriate actions to take. First, you have to build extremely deep understanding of the models and the ways in which that risk might obtain. And so one of the big things that we're focused on, the Security Institute is, yes, looking at cyber risks and yes, looking at chem bio biological weapon and chemical weapon construction risk. But at the heart of it, to also look at loss of control and alignment risks, which are really the ones that relate to the existential element that you're describing.

And so we have been able to be, as I as you know very well, the only lab in the world to be able to evaluate models on each of those risks. And in doing so, both support the companies to safeguard models from those risks, but also internally share that understanding with both national security agencies and the Cyber Security Centre to be able to mitigate those for critical national infrastructure. That's one part of it. A second part of it is that the reality of where we are, as, as you may have heard me say, is that to me, this is the central question for both our domestic and national security.

The fact of large scale firms now being primarily exposed to cyber risk from AI. The fact that warfare, whether in Ukraine or Iran, is primarily determined by the quality of your drone and counter drone operating systems, increasingly underpinned by AI, those facts mean that on your question of what is a proportionate response, to me, the most proportionate response would be for Britain to actually measure up to the scale of the diagnosis, the scale of investment in AI and a centrality of it in our public conversation. And then the third thing to say is that given the nature of how models are being developed, which is not just closed licensed models in the United States, but increasingly openly proliferated models in China over which we have no statutory ambit, means that we can't simply rely on assessing the models.

We have to rely on societal hardening and resilience much more broadly, and that requires a whole of government, but whole of society approach. And there are a series of things that we've tried to do through the National Security Council to be able to prompt that as well.
16:50
16:51
16:51
16:51
16:52
16:53
16:53
16:53
16:54
16:54
16:54
16:55
16:55
16:55
None Portrait Speaker 13
-

Okay.
16:55
16:56
16:56
16:57
16:57
16:57
None Portrait Speaker 21
-

Very happy to.
16:57
16:57
16:58
16:58
16:59
16:59
16:59
17:00
17:00
17:01
None Portrait Speaker 18
-

I was perfectly.
17:01
None Portrait Speaker 13
-

Primed for this.
17:01
17:01
17:01
17:02
17:02
17:02
17:02
17:02
None Portrait Speaker 13
-

Chris.
17:02
17:03
None Portrait Speaker 21
-

Thanks, Chris. And, you know, it brings back to our my wider approach of what this government is hoping to achieve with our make work pay agenda, that we fundamentally believe that you cannot grow the economy with people in insecure work. And that applies whether it comes to technology, to contracts, to so much more that we are obviously working on, um, in the department, I suppose when we look at AI in particular, in terms of the impacts that we're seeing, and I can completely understand when people see BBC news notifications about X amount of jobs.

Research report has done this. You know, what does it mean for me in my job and in my role in the future?And how can I best prepare my career and my skill base to adapt to that and to respond?And how will my employer support me?Um, so I suppose when we look at the, um, evidence to date around AI specifically, um, in terms of its reducing employment, employment, there is at the moment quite inconclusive evidence to exactly that particular point that AI equals, um, a significant reduction in employment.

Um, you know, it's obviously a lot of wider instances where an employer might make certain decisions about their workforce, but where AI is a particular factor where, um, some employers might say that, yes, okay. We're introducing different types of technologies to increase, increase efficiencies or to change how we do things. Um, I think the only way that we get that, right, as I keep talking about is that, yes, okay, we want to make changes because we know that if done well, we can be more productive workplace, we can absolutely grow our business, compete internationally, that the people that ultimately they will be using that your workforce.

So you want to be able to for it to work. You want them to feel that it is fair, that it is transparent in terms of that process, and that it will ultimately deliver its objectives, which I would hope the business would want to achieve. So, um, I suppose while the evidence at the moment is quite inconclusive about the potential impacts, for me, it's about if we don't get it right, then it's the quality of work for people that's impacted and follow up on future.
17:04
17:04
17:05
None Portrait Speaker 19
-

If I just follow up, I think, um, the situation we're in is exactly as Kate says, is that, um, there are good reasons for people to be worried based on sort of micro evidence, which is to say the scale and speed of AI as an economic force has been very unusual, I think, in 2023, it as a sector AI took something like 180 days to add $1 billion of revenue. Um, now AI as a sector is taking less than two days to add $1 billion of revenue. The scale of that uplift is exceptional.

And then the second micro evidence is that when people have done theoretical work to understand the exposure of tasks to AI, they have found that the models capability are increasing so much, starting with language to coding, to data insights, and increasingly to mathematical calculations, that the scope of possible exposure theoretically is very high. So that's the micro evidence. And against that you have the macro evidence, which, as Kate says, effectively shows, at least in the UK, no signs of net job loss and no systematic patterns of bias in terms of either age or gender or geography.

What are theoretical reasons, then to believe that we have some view of what might be the answer to your question?This feels very unlike some past technological advances, in that if you just take those task exposure measures, it feels like last previous technological revolutions were, on a relative basis, better for higher paid workers, knowledge intensive workers, and they were better for women, and they were better in places of agglomeration, often in cities. Um, theoretically, again, very important to say empirically, we don't have good evidence at all. We don't have evidence on these impacts.

But theoretically, if you look at task exposure to AI, um, highly paid jobs. Um, female heavy and knowledge sector white collar heavy. Those are the ones from a task bundle point of view that appear to be theoretically most exposed to AI.
17:07
17:07
17:08
17:08
17:09
17:09
17:09
17:09
17:09
17:10
None Portrait Speaker 13
-

Right.
17:10
17:10
17:11
17:11
This debate has concluded
Return to start of debate - Return to top of page