Kanishka Narayan | Future of AI, From the Inside

0:00When you want your spring break to feel like... And your kid's pool day to feel like... And your hotel bed to feel like... Ooh, and room service to feel like... Because at Hilton, hospitality feels like... Your cabana's ready. Would you like fresh towels? It matters where you stay. Book now at Hilton.com. Hilton. For this day. This episode is brought to you by Indeed. Stop waiting around for the perfect candidate. Instead, use Indeed Sponsored Jobs to find the right people with the right skills fast. It's a simple way to make sure your listing is the first candidate seat. According to Indeed data, Sponsored Jobs have four times more applicants than non-sponsored jobs.

0:49So go build your dream team today with Indeed. Get a$75 sponsored job credit at Indeed.com slash podcast. Terms and conditions apply. Welcome to this episode of Jimmy's Jobs of the Future. I'm delighted to be joined by the AI Minister today, Kanishka. He's somebody that I've wanted to have on the show for a little while since he was first elected in the 2024 general election. He's a rising star on Labour's ranks and he's got a pretty incredible backstory having arrived in the UK age 12, getting a scholarship to Eton, studying at Oxford University and then Stanford as well. He's had a remarkable career so far and it's great to see him in the AI brief and chat to him where he thinks the future's heading.

1:33On to today's episode. Make sure you subscribe so we can continue to get bigger and bigger guests.

1:45Kanishka, welcome to Jimmy's Jobs of the Future. Great to be here. Now I was going back through your history and you arrived in the UK age 12 what was that like? I look back on it as a absolutely wonderful experience we landed at Heathrow we were going to Cardiff and we got the late night coach from Heathrow to Cardiff and I remember arriving in September this is over 25 years ago so So arriving in September, it was raining in South Wales. It was cold. But I remember stepping out and seeing the sort of autumnal colours around Cardiff. We went to Richmond Road where we stayed overnight with some family friends.

2:28And it was completely magical. It was magical visually compared to what I'd experienced before. But it was magical because the thing that had attracted me and my family to move was this huge sense of opportunity and excitement about what Britain meant. What did your parents do? So my parents trained as lawyers in India, but my dad came here to study postgraduate work in business. And for about three to five years, they ended up working in a range of kind of a regular job. So they did agency work in hospitality, hotels. They went around late night when shops shut. They went around auditing the shelves to make sure that the supermarkets had an audit of the stock inventory as well.

3:16So they did a bunch of agency precarious work as so many people do before they find their feet and ultimately re-qualified and ended up setting up a law firm together as well five years into our move here. Interesting. And why was it Wales that you were heading towards? Yeah. You know, I think I look back on it. My mom had spent some time in Wales before she got married. She'd studied there. She had some family connections. And that was a really interesting thing, because the thing that I only later found out about Wales is so many people feel that it's the sort of size where you can come in, get a warm welcome, and then move on together, because the community element is a really tight end thing.

3:57People say in Wales, you're always only one degree of separation away from everyone else in Wales. That kind of holds true. And so I think when my parents were thinking, how do we give our family the best chance or opportunity? Wales really stood out for the fact that, yes, there was familiarity, but there was this belief that that familiarity would mean a warm welcome and a helpful step up. And then a year later, you win a scholarship to Eton. Yeah, it was a very odd experience. You know, in India, you kind of look at Britain and you say, what are schools that are meant to be good schools? And I remember Eden had been covered in the news.

4:30And so I came here with no idea what it meant in the kind of broader social context. So my brother and I used to walk up all the time to Kattes High School and Inner City State School in Cardiff, which is where we studied. But in my mind, I thought this might be an exciting thing. And I had this brilliant Welsh teacher, Mr. Thomas. and I remember he sat me down and he said you know you seem to be really geeky and excited about studying math and science and I think I think you should apply to apply to a place like Eton and so I remember getting my head teacher at Kitej to sign the form he was on the bus breaking up a fight and I kind of walked up to him and I said I'm going to apply for a scholarship exam at Eton and I need you to sign this form and he kind of was just had like no cognitive space to think about signing a form so he just sort of scribbled something on the form I sent it off and you know had a slightly patchy sense I was good at maths and science not so good at sort of English history at the time and so they were still generous enough to give me a huge amount of financial assistance and totally changed the trajectory of my life completely can you remember much about the exam what's an entrance exam to each of the life I mean the maths stuff is it's a bit like the kind of math olympiad so it's not your usual curriculum you actually I think are not supposed to finish the exam the whole point is they're testing not that you can get 100 out of 100 they're testing whether given the problem that you haven't seen before how do you sort of engage on it and so the maths bit was really engaging and challenging the general paper was kind of IQ test like so logical puzzles things like that and then the history of the geography you know some of the other stuff was just really tough for me because it was all new stuff and how did you find your time then at Eton because it's got it's got this sort of from the outside quite an interesting reputation and perhaps not always you know the best over the last couple of decades in terms of its impact on UK politics and so on how did you find the time there yeah I think the impact thing is interesting you know from my point of view as I said my parents were working in minimum wage jobs very precariously in petrol stations and hotels my brother was at Kotez high school we used to live in a one-bed flat so my brother and I shared a room we never had a room and then I was eating I had this whole room to myself this huge expansive educational experience as I said it totally changed the trajectory of what I thought I could do in life so for me it was a huge gift but in terms of impact you know I feel like for some people an experience of that extreme privilege educationally can instill huge conviction and a sense that you've you deserve to be running things and that's one path.

7:13The other path is that there are lots of peers of mine as well who come out and say that they recognize that privilege and they see that that isn't available widely and part of the reason it's not available widely is because of the context of our society and institutions and they want to use everything they've gained from it to try and move the needle on that question. And, you know, I feel part of one of many people who came out of Eton who want to try and make a mark on that. And I think that story is not a frequently told story, but I think it's certainly my story and a story that I feel very, very keenly.

7:51And what is it about the school that really does encourage people to world public service, do you think? Like, because I think it's actually one of the challenges that we have as a country is that sometimes public service, being a civil servant politician, isn't perhaps held in the same status as it is in Singapore perhaps and so on but what is it about Eton that encourages people down that path? Yeah I think from my point of view I turned up very interested in wanting to become a scientist I wanted to work at kind of NASA or something and really push the frontier of understanding scientific truth and left thinking that I wanted to really work on public policy, changing the course of how the country and the world could be better through that.

8:35And I think it's because you're encouraged to think big, you're encouraged to be engaged in the battle of ideas. And especially in that moment, you know, I left school in 2008. In that moment, if you asked people, what is the biggest way in which you can make a positive difference in the world? People would say public policy at the time, economics, development economics was a big thing as well. And generally, people felt that if you really wanted the country to be better, that politics was the way to do it. And so I think it's a thing that is attractive to people who want to make a positive difference at scale and try and do something deeply impactful to consider that politics and public policy has for decades, if not centuries, been the primary way of doing it.

9:18And I think will remain so in the years ahead as well. At the margin, that changes from time to time. Sometimes people want to go into finance because it feels like the tough and interesting thing sometimes people want to do technology as of now but I think the common core in Britain that I think is really distinct is that people believe that public policy the ability for us to get together and through accountable means of democratic representation change the country for the better I think that's a really long-standing tradition in some of our schools across the country and you joined the Labour Party at 16 did it feel different like obviously Eton has produced a lot of conservative prime ministers but not as known for producing sort of labor politicians how was that experience one of the things i feel strongly is that we massively overdo the impact of institutions on people's beliefs uh in this in this country um so for me a really foundational experience was that every summer i'd go home and my brother and i you know would share this room and live in kate's and uh and in sixth form we worked in a kebab shop overnight and so i got a pretty good sense of what South Wales was experiencing in that part of the community.

10:27And then I got a sense of what my friends from Eton were experiencing. And when you experience that, you want to make sure that kids in both sets of contexts experience the best shot at opportunity. And when you think that, and you look at history, the fact that, you know, Healy and then Brown and then now Rachel Reeves are all the best occupants of the role of the Treasury from a fundamental focus on shifting the needle on child poverty, you feel like this is a great vehicle to be able to make positive change in the country. Other people take a different view because they've thought it through.

11:00Now, I will say that there is a tradition of people from all sorts of schools, including Eton in the Labour Party. In fact, I think there's been a serving member of the Labour Party in Parliament from Eton for every year up until 2010. And so the gap is an aberration uh rather than uh rather than the norm well that's a fact we didn't know um and was that your first job then working in the kebab shop overnight was that the first thing you did for money yeah so i think uh i are i i think i can't remember if i was the kebab shop was first or i delivered the metro newspaper on saint mary's road in the morning in the winter it was bitterly cold and i so viscerally remember it my gloves were really really not great gloves and i was um handing out metro newspapers on uh on saint mary's road when commuters were going through at about 7am it was freezing cold that's interesting that's often the newspaper i sometimes think of this as generational in terms of like boomers and gen x often their first money will have been from newspaper deliveries the sort of next generation our generation millennials tends to be kind of like hospitality bar work etc and now actually the younger generations and the first money online and i just think it is such a formative experience it perhaps has more impact on the way that you see the world than institutions, perhaps, given what you were saying earlier.

12:17So let's fast forward a bit. We'll sort of skip out Oxford and Stanford. Maybe we'll come back to those one day. But you get appointed the AI minister by Keir Starmer. What was your reaction when you got that phone call? Well, the chief web called me and I was in a cab going home. and for a period of time both pre-election and then immediately after election I was off the conviction that the primary crisis in Britain was a crisis of prosperity that the best path to fairness was a path that worked through greater prosperity a richer country was going to be a fairer country and that AI and technology had huge opportunity to be the central driver of that prosperity so when I got the call and was offered the role I just felt this huge sense that I'd spent so much time thinking about it, preparing for what some of the public policy answers were for Britain when it came to technology, that the opportunity to try and then do that in practice was a huge gift.

13:15And so I felt very grateful. But more than anything else, I remember going home, I wrote down a list of sort of before I got into the institution and became native, I thought, what are the 10, 15 things that I think I want to try and do within the first three months that will move the needle? and it's just very gratifying. I remember writing my first note to the Secretary of State and saying there are 15 things that I think we need to try and get done within the next 90 days. I think 12 of them we've got and done. So I just feel really grateful for the opportunity. I think the driver of prosperity and growth for our country is going to be AI for the course of the foreseeable years and at the same time there are lots of difficult social questions that are also fundamentally driven by AI and the experience of safety online.

14:04And so the combination of those two things, I think are the two most important things in British politics. So what is the job of the AI minister? Because it's quite a new job in some ways. There aren't that many holders of it previously. So what is the job of the AI minister? I think about it in terms of outcomes. There are two things that I need to make sure that I deliver. One is to make the country richer as a result of capturing all the opportunities of AI right across the country. So drive growth, but drive growth that everyone can be a part of. And then the second part of it is that alongside that, to do it not just at speed and scale, but to turn the steering wheel in the direction of our values.

14:39And that to me is really fundamentally about making sure people are safe online, that our adoption of AI is responsible, and that we're driving it with British values as well. So yes, go fast and go further in terms of growth as a consequence of AI, but to also give it direction in the direction of the things that are really important to us in Britain. How do you define AGI? Wow. So AGI is, there's a narrow way of thinking about it, which is what is a capability for AI models which can outcompete or match humans in terms of core tasks that are critical to intelligence. So language, reasoning, vision, memory, some of the fundamental building blocks of intelligence.

15:26At what point does AI meet the threshold of human competence across a range of these tasks? So that gets you AGI. There's sort of two deeper meanings to it, which are less to do with just what a model can do. But if you think about what the purpose of a definition is, the purpose of a definition is to try and understand when something will make a difference in the real world. And so, yes, the model capability you define in this way. The second way I think about it is at what point, not just because of the model, but the scaffolding and the tasks around agency, at what point do those models combine to actually have human-like capabilities?

16:04Because at the moment, the kind of AI is sort of like a team member who will come and just do one task and then will drop out and then will come and do one task and drop out. That's not a very useful human-like intelligence capability. the capability of not just vision and reasoning, but the ability to knit those together. So the second thing I think about when I think about useful ways of thinking about AGI is when does the sort of knitting work out so that models can work together and develop genuine, agentic outcomes? And then ultimately, the question is, when do they not just have the capability to do that, but the economics of it and its integration with how we live our lives becomes such that AI truly becomes deployable in a way that you can do that at your home, you can do that in a robotic supermarket or manufacturing context, you can do that in office.

16:52So those are the three steps. I think a lot of the conversation is focused on the first, which is when does the model become so good that it can match human capability across a key range of specific capabilities. But for me, the big picture is to go from that to then it having a degree of agency and then having agency in an economic way. Yes, I think the unproactive point is an interesting one actually it's like having a highly qualified master's student who is is quite unproactive with it um and do you think that we will be there by the time of the next election because i've asked both peter kyle and the prime minister this and it's been interesting to get their responses but yeah do you do you think we'll have agi by the next election I think on this first rung of the ladder, which is, do I think we'll get models that can match human capability on these range of tasks by the end of this decade?

17:46I think the answer is actually that's a pretty highly likely thing. So whether it's 2029 or whether it's 2032, somewhere in that range, I think it's got a high likelihood. But I don't think it's a slam dunk. I think it's dependent in part on some of the key bottlenecks of energy and land and being able to build the infrastructure for it and memory in particular as well. But if you just said to me, what's your bet for the timeline? I'd say 50, 60 % chance it happens by the end of the decade. All right, interesting. That almost makes you the most sceptical of the three that I've asked. Is that right?

18:22Yeah, well, Peter Carr was like, yeah, I think it's going to be there. and I was literally interviewing Keir Starmer a couple of weeks later. I said, well, here's your Secretary of State. And he just said, well, I think Peter's pretty close, but he is an ambitious guy. And then he moved him. I can't be the doomer of the Prime Minister, the Secretary of State for Business and the AI Minister. The AI Minister needs to be the hype man for AI. I think there's a good chance it can be by the end of the decade. I think that's a fair summary of it. And what do you think the impact could be, right? Like that's the critical thing that we're talking about here.

19:00Obviously, the impact on jobs. What do you think will, you know, what's the job that's most at risk over the next three or four years? So I think the reason I'm really keen on making sure we talk about each of these rungs of the ladder, not just what the model can do, but then is the stitching in place for it to become genuinely authentic? And then beyond that, is it at an economic level so that we can actually deploy this at scale? Those two steps are really important. You know, on the model, as you said, maybe we're five years out, maybe we're seven years out. But I think the next two steps actually will take a long time because as with so many foundational technologies historically, it's not just whether the technology is good enough.

19:39You're thinking about the steam engine. You think about even things like ATMs. I think for at least two decades after ATMs, the number of accountants in the world grew to decline. And so the reason for those is because is the patchiness of capability and the fact that our organizations, all of us take quite a long time to reform ourselves so that the technology works with us and for us means that there's a way to go. So that's the starting point. On which sorts of jobs are most exposed? I think the jobs that appear to be most exposed are jobs that involve one sort of task or narrow set of tasks where the outcome can be validated so that a model, you can tell a model, here's some data, try a task.

20:19It tries the task. you can validate whether the outcome was good or bad, and you put that back into the training, and the model improves. So tasks that are narrowly specified don't involve lots of different sorts of things knitted together and where you can validate the outcome. So I'm thinking, for example, exam tests and things like that, written exam tests. Clearly, those have been done well. A lot of maths, actually, and scientific progress is being driven by AI. You can clearly see that administrative roles, which largely involve mechanical writing and copying of particular things. I think those sorts of roles are most exposed to AI.

20:55But I think there are so many roles in the world which involve not just the tasks, but the knitting that stitches it all together. And I think when it comes to those sorts of things, I think there we have a bit of a longer runway to go. And you've got this sort of announcement today about sort of upskilling people. that's right because the biggest question jimmy is um in terms of what the impact of ai is on jobs is uh to impress on people that we have agency on this question it's not just something that we sit back and say when do you think aji comes and kind of what what happens as a result no we are participants in this we can shape this and so the announcements to this week have been about shaping it in the direction of human progress and so we had committed to train seven and a half million people by 2030 in AI skills.

21:44We are well on course to do that. In fact, we're going to breach that target and be even more ambitious. So we're going to go punch bigger in that context as well. We've got an AI hub. I encourage everyone will share the link with you. We're encouraging everyone to go online. You can get essential courses so that you can upskill yourself in this context as well. And we're setting up local programs, tech local. We're putting 27 million into it so that kids with placements in terms of work experience, graduate jobs, local jobs across the country, the opportunity of AI becomes way more spread than the opportunity of technology was in the last two decades where I felt it was too concentrated in just London or Silicon Valley.

22:22This time we want to spread it right across the country. And how do you think about reaching the people that need this most, right? Yes, we'll put a link in the show notes to it and so on. But I dare say the people that are already watching this show are pretty engaged in this stuff right like how do you think about reaching you know all across the country in places that you know might not have access to this as much i think this is a really critical thing so you know the first thing i will say by the way is online distribution has meant that you will be surprised at which parts of the country people might be engaged with you on and so there are kids who came to me from my constituency in barry island and in south wales who said they had seen venture podcasts online and as a result wanted to build a career in tech and starting a company.

23:10The context of imagination for kids in this country, right across the country, has really shifted because they can now engage with content right across the world. And so Jimmy's jobs, I'm sure, is going to be played in South Wales and there'll be kids who'll be looking at AI skills as well. But broadly speaking, one of the things that we're really keen on doing is making sure exactly as you say, that we're finding other ways of reaching people who we really want to be a fundamental part of the economic opportunity here. There are two things that are really on my mind. Are you a fan of Jimmy's Jobs of the Future?

23:39Hi, I'm Sonny, one of the producers behind the show. And we don't just make this podcast, we produce some of the best business content out there, working with global brands and founders. Over the past year, we've flown around the world and demand keeps growing. And with four prime ministers trusting us with their appearances on this show, you're in safe hands. So if you're already in the market for a podcast or standout business content, why not start with us? Drop me a line at sunny at boxlight.io, try our podcast calculator or book a meeting at boxlight.io. That's boxlight.io. Now back to Jimmy.

24:14One, we've got a series of AI ambassadors that we've engaged, Tom Blomfield from Monzo and GoCardless fame being one. and my big ask of them is to work with us to reach schools and workplaces right across the country to be able to take this message out and say we think kids growing up in every part of the UK Scotland, Northern Ireland, England, Wales are all able to now make the most of the opportunity so we're going to get the most engaging people, people who've done it to go and say that you can also do it regardless of where you're growing up and the second part is that we want to make sure that the power of public services which are spread across the country are being made the most of in spreading the message and so one of the big things that we've announced recently is that we're going to take some big moves in education in particular to make sure that teachers are supported with AI that learning can be personalized so that kids and teachers right across the country can see the opportunity of AI directly and as a result really feel like they can drive that progress regardless or where they grow up what have you used personally to sort of train yourself on ai as the ai minister uh i have a couple of different things so one you know the usual uh chatbots i've uh recently switched i used to mostly use chat gpt and claude i've now actually felt the gemini is probably my most frequently used uh chatbot i just feel the range is great the deep thinking research stuff is really helpful.

25:41I have done some very minor coding on Cursor and so I find being able to go on a chatbot like Gemini increasingly or Claude Code and then be able to sort of use Cursor has been interesting although I've mostly worked with others in doing that as well. And then the third thing is I use a couple of the tools Opus Clip and Veed in particular for editing and video and kind of some creative content work from time to time as well. I'm kind of pretty amateur at it. So your producers are going to be telling me how bad a job I do compared to their remarkable skill. But those are three things that I feel are most useful.

26:20We will get you some clips for it. But what do you think? Because it's funny sometimes friends ask me when they're moving jobs or whatever, like I've got a bit of time. And I just say to them, well, look, get a pro account of pretty much any one of them and just play with it and tell it what you're interested in etc but what do you think will be the foundational skills because obviously it still matters what people are putting in at that stage right so what are the foundational skills that you think are going to matter in the future one of the things that i feel even just for myself when i'm using it is that the vigilance i have to have about the quality of information has to be way higher than when i'm engaging with someone in person because when you're engaging with someone in person you have some proxies for where you're at right if it's someone who's if you're engaging with a kid and you're asking him about something specific you know that there might be creativity but you don't want to believe all the facts that might come out of it if you're engaging with an expert with some kind of proxies and a sense of reputation for quality you can kind of you've got to be vigilant still but you can you know where you stand whereas with The AI chatbots, I feel like the fundamental thing is how you retain the cognitive vigilance to say, hey, I'm going to call bullshit on this and on other bits.

27:40I'm going to kind of feel like, hey, this is the right information. That is really fundamental. And so how we build that, especially in a context where we rely more and more on the output of the chatbot, how we make sure that the foundational bits of analytical understanding, cognitive vigilance generally, so that ultimately we are all becoming exceptional managers of AI systems where AI can do this and that for us. but we become the managers who are able to test where it's going right where it's going wrong and how we want to use it in the day-to-day context that we're working on i think that to me is the most foundational and fundamental thing how can young people build that cognitive vigilance because i think it's a really interesting phrase but it's one of those things that i think yeah it's easy for us to do when you've got a bit of experience of the world and so on but as a young person how do you build that you know i feel one of the things in this context is uh we've got to remain very focused on how the education system i think in particular can build some of the things that are really building blocks for it so one making sure that kids are being able to uh engage with when they're doing an essay yes they might use chat dbg to come and write the essay but then you come and you get them to debate the essay with you one of the things i found early on in my career and educational context was you could write whatever you wanted, but when someone said, explain this to me and run me through all the steps of logic, and you got beaten up for it because people caught you out when you didn't really fully know what you had written.

29:16I think that sort of thing of really understanding each step of logic and having that tested in verbal form or forms that you can't just hide behind an essay, I think that's sort of critical. I think even when you're writing essays and you can't do that verbal stuff, being able to test for originality of thought. There are some bits that are still not easy to do with AI. I think there might be small ways in which you can do it through the assessment process, but for me the fundamental thing is how we get kids to engage with each other, but in the classroom context, in building blocks of logic when they're writing things or when they're thinking about maths, how they think about really the logic that allows you to do calculations that you're doing.

30:00I think we need our assessment framework to more and more focus on those building blocks, the sort of underlying stuff, rather than just the answer. Yeah, I think that's... That verbal jousting, I think, will become more and more kind of like important, actually, and how we test that, because we have mainly tested people through written word, really, and that's going to have to change. But you're not in education yet, Do you think government, and I'm interested in your experiences as a relatively new MP and as a relatively new government minister, do you think that government sort of over-indexes on risk when it comes to new frontier technology?

30:49That's interesting. I think one thing I do feel that's happened over the course of the last couple of decades is that we haven't indexed enough on the upside of technology for sharing prosperity across the country. Post the global financial crisis, you can clearly see that America grew way faster than us and was way more productive than us. And one of the biggest, and that's had huge benefits in terms of prosperity you can spread across the country, it's had huge benefits in terms of global standing, sovereignty, national security, hard power. What happened in Britain in the last two decades was broadly that we massively under-invested in technology and capital investment in technology.

31:27And as a result in part of that, we have not been as prosperous as we ought to have been. And so if you just take that and you say, if that's one of the primary crises in this country, then you've got to flip that. And we've got to capture the opportunity of technology this time around, and AI in particular, because it's going to be a huge driver of prosperity. and that prosperity will be a driver of both fairness and of national security. And so, yes, I think we need to look at the upside a lot more. I think that said, one of the things I find is that the conversation on risk is not necessarily over-indexed in risk, but patchily indexed on risk.

32:01What I mean by that is there are some things in risk which matter. People's experience of technology is driven by whether they can trust it, whether their kids feel safe from it, whether they feel like it's not going to cause bits of bias. But when we're even thinking about that, I think some of our policy debate ends up being behind on where the product is. So recently we passed a law which said we should now allow enforcement authorities like the police and some trusted authorities to use AI to spot child abuse material. Because the law didn't even allow that. The law said the technology is bad.