85. The most important Industrial Revolution in history

0:10Hello and welcome to The Rest is Money with me, Stephen McGovern. And with me, Robert Peston. Now, today we're joined by someone I've known and admired for a while. He's Azeem Azhar. He began his career as a journalist, as all the best people do, for The Guardian and The Economist before moving into strategy at the BBC and then becoming a serial entrepreneur in technology, largely. And it's because he's one of the most insightful analysts and pundits on how tech is changing the economy and our lives that we want to talk to him today. Yes, because Azeem advises politicians and business leaders on the challenges and also the opportunities as well, presented by AI and Emerging Technologies.

0:57He's got his recent book out as well, like all the best people do, Robert, Exponential. So we've got loads we want to get into with him. So here's our interview with Azeem Pazar. Azeem, absolutely great to see you. You've been an observer of Elon Musk for some years. I think some of us are struggling to understand him, certainly in terms of his approach to business, maybe as a human being. But anyway, we're hoping you'll enlighten us a bit about the real Elon Musk. But I think we kick off with one of our great obsessions on this podcast, which is artificial intelligence, the industrial revolution that we're living through as a result of artificial intelligence.

1:40I just really wanted to start by asking you, what is your current take about the significance and pace of this industrial revolution caused by artificial intelligence? Thank you so much, Robert and Steph. And I know from our earlier conversation that both of you are real people and not synthetically generated AI avatars. So that maybe tells us a little bit as where we are right now. It's such an unseemly debate at the moment. We had one side of the boosters and the other side of the sceptics. And there seems to be there's no twain where where the two of them meet. And what I think we can say is that since November 2022, when ChatGPT rolled out rather unassumingly, it has electrified the technology industry in Silicon Valley at first and then elsewhere.

2:31And that electrification is not really just about the really big companies like Google and Microsoft. it's really about the way the entrepreneurs and the venture capitalists are starting to think because they are the ones who build the products that become quotidian in five to seven years time. And Silicon Valley turned on a sixpence around November 2022 because they saw the potential for a new technology platform, one that could launch an industrial revolution. Now, the mismatch happens in that building things takes a lot more time than getting excited about the potential of something being built.

3:10And so what you start to see is capital flow into these entrepreneurs from venture capitalists who deploy risk-oriented capital specifically for new innovations, that being followed by speculative capital, less committed capital. And it's that speculative capital that starts to give rise to exuberance. And then everyone gets into a mode of needing to effectively big up what it is they're doing. But they're doing it far faster than either developers can develop really excellent products or company managers can figure out what to do with those products in their firms. I think this is an unavoidable cycle, to be honest.

3:47We saw it with the Internet. We saw it with the telecoms boom at the late 1990s, early 2000s. And for those of us who are super old, like sea turtles, we would have seen it with the railway boom in England in the 1850s. So is this a case of zine then of it's just over excitement? You know, it's not that it's the emperor's new clothes where all of this kind of interest is going to come to nothing. It is going to happen, but it's just over excitement at the start of it all. There's a lot of overexcitement, but this is very different to other speculative bubbles that we know from history. Tulips with Tulip Mania didn't do anything useful.

4:25NFTs, non-fungible tokens, those ways of creating digital art haven't yet done anything useful. There was just speculation. But what I see when I talk to bosses of large companies and when I actually use AI tools today, I see them doing very, very useful things. And so there is a sense that there's going to be a lot of utility there. I think we have to understand that there's a bit of crowd psychology that is going on. People don't want to be the ones who miss out. Bosses in companies today were young. They were in their late 20s and early 30s when the internet came around in the late 1990s. And they probably saw their company struggle in the face of new entrants.

5:11And these CEOs today have all been schooled in the world of digital and mobile and e-commerce. So they understand the potential of technology in the way that perhaps their bosses 30 years ago didn't. And so I think that creates much more of a stance to wanting to use this technology, partly because of their own personal experience. And that crowd psychology ends up a bit like watching five-year-olds play football, which is, you know, we're all going to chase the ball at the same time. But it's undeniable that there's a ball on the pitch and there's undeniable that someone ought to try to kick it into the goal.

5:43The thing that I find really interesting about this, Zima, is you can really tangibly see how it can help business. But I think there's like a communication issue in terms of people in what they would see as more traditional jobs like teaching or in care and things who might think, hang on a minute, this isn't going to help me. But yet, for example, in teaching now, you can get all this kind of generative AI stuff that will create lesson plans for people. But I don't think a lot of that's communicated to people very well, is it? So you get that sense of fear about AI. you know it's going to take your job soon it'll be robot teachers but actually it's not about that it's about doing all that hard graph that makes the reality of teaching in the classroom much more difficult because you're stuck doing all of this work that AI could do.

6:31Well I think the technology industry has done a really poor job at communicating and more importantly listening to people. Instead we've been the recipients of proselytization and so many of these promises have been made time and time again, and they haven't been achieved, whether that was with CD-ROMs back in the day, through to interactive apps, you know, the famous Quizlet app, none of these things have really had an impact on the quality of people's day-to-day work, at least as they feel it. And that's really where all of this starts. So I think the technology industry has done a very, very poor job of explaining it.

7:07And that's been amplified by many parts of the media where it's much, much more interesting to talk about robots taking every single job or helping us colonise Mars than it is to say, here's a robot system that could reduce the time it takes for doctors to fulfil their regulatory reporting requirements. I mean, that's really valuable because we want doctors to have more time with patients doing less paperwork, but that will never make a news headline. And that isn't something that the largest technology companies feel will will change their business if they talk about things like that. We've talked a lot about the positives that can come from AI.

7:45And I think Robert and I are both very pro-artificial intelligence and the way it can improve our lives. But obviously, there's a lot of people worried about it. And things like the AI's role in disinformation, the fact that generative AI can be used to create fake things or misleading things and how that can impact, for example, elections or how it can stir up hate on social media and we've seen that play out terribly over the last few weeks is this something that worries you or again are you an optimist about that sorting itself out my answer today is quite different to where it would have been a month ago to that question so my answer today which is after these you know extreme far-right riots has really drawn attention to where we currently sit with this very, very odd situation where we have one of the big communications platform, which is X, formerly Twitter, entirely owned 100 % by a single individual who doesn't have to report to any other shareholders other than himself in the mirror in the morning.

8:51And I think that changes the nature of the conversation because a lot of the assumptions around how we tackle misinformation, disinformation, were based on the idea that there were a few large companies, these companies were probably too powerful for our good and for the general public good, but at least they had their own risk management, they reported to their own shareholders, they would be willing to show up to Congress or to the House of Parliament or respond to regulatory requests. And I think what I saw over the past month, where the owner of X was willing to ignore all of those, made me think that perhaps we've got quite a knotty problem there.

9:34And just to be clear, though, we're talking about Elon Musk and X, are we seeing, do you think, in whipping up the hate and chaos that was then manifest in those riots? Do you think we now have any evidence that AI itself is being abused in terms of the misinformation? Or are we still in the territory of, in a sense, human beings? They may be in Russia, for example, but are we still, do you think, in the territory of human beings sowing the confusion? Well, when we look at misinformation and disinformation and its effects, what it seems to have done is appeal largely to people whose minds were already made up.

10:25And it tends not to appeal to the minds of people who take a different view. And I think this is really relevant in the context of the upcoming American elections, where there are one in seven, about 14 % of voters identify as swing voters. And the question is, can misinformation or disinformation of any type really, really affects the outcomes. Now, we haven't really seen yet much AI generated material that I think could change people who sit in the swing category. They're sticky, I think is what you're saying, because I mean, we've all seen fake videos, but they become sort of events in themselves, they get exposed as fake.

11:06And even though during there's a period of a few hours, where they may be influencing what people think, you tend to get enough focus on it for people to recognise, most people to recognise that this is fiction. Right. I think that's exactly right. And the academics who looked at the 2016 and 2020 elections in the US struggle to find really strong causal pathways between fake material produced by humans at that point and sort of very limited generative systems and actual changes in how people voted. What those things tended to do was appeal already to people who are extreme partisans. The one thing that it can do is it can impact the overall sense of trust that we have in the system.

11:51And that's particularly appealing if you're an adversary. And what happens when we don't have trust in the system is we don't feel we need to vote. We might doubt the processes and institutions themselves. And to quote Steve Bannon, if you just flood the zone with this material, you can certainly get people doubting everything that we see. But I don't think at this stage, we have seen very many examples of that, that as you say, Robert, are very sticky. But it is still a subtle and important point that you're making, though. You know, one of the things I've been concerned about since July the 4th is how low the turnout was.

12:2940 % of British people didn't vote. I saw a bit of analysis the other day by one of the polling companies that said that if that 40 % non-vote was seen as a vote for the so-called apathy party, the apathy party would have had something like 95 % of all seats in parliament under our first past the post voting system. And it certainly is the case that in a world where trust has collapsed, in a world of confusion, a lot of people just think, well, I don't really know what anybody stands for. Therefore, or why should I bother to vote? And so in that sense, if AI is deployed by bad actors, Russia, for example, simply to undermine confidence in the system, that is dangerous.

13:18That is worrying. It is dangerous and it is worrying, but we have to also understand how that material gets in front of the eyeballs of people who might be considering joining the apathy party or not joining the apathy party by default. And of course, that means that you are dependent on essentially the social networks for the distribution of that material. So it's one thing to see an example of somebody producing a fake video of Keir Starmer saying something, but it's another thing for that to go all the way through, get a lot of distribution and then be believed by enough people, especially considering the types of activities that you've talked about, sort of debunking that goes on, the media discussion that goes on.

14:02And the platforms all work quite differently. You know, Facebook is different to TikTok, is different to X, in terms of how they amplify and distribute that material. And I think the riots that we just saw showed that example, because it was really, really dependent on motivated individuals to amplify the first false tweets about the murders in Southport for this to then get algorithmically picked up and driven. The thing that I'm interested in that you touched upon, and I know we're going to talk about Elon Musk specifically, is that ownership point. Because we've talked a lot on this podcast about how there's this whole movement, particularly in the tech world of this effective altruism and this idea that all of these people are running these companies and doing this great AI for the greater good of the world but yet we saw it all play out when Sam Altman at Open AI the board of questioning whether he really was an effective altruist and you know then he left for a bit and then came back and I would love your take on that whole thing of the ownership of these companies because as you rightly explained the fact that one person owns X what that means in the AI world as well and whether this effective altruism is real and what it all means for us.

15:23And I mean, there's a related point as well in the case of chat GPT. It's what we understand to be the case about what happens to that ownership at the point that you get to artificial general intelligence, the point of singularity where artificial intelligence can do everything we can do only better. On the outside, it looks extraordinary that Microsoft can have such apparent power over that business only for that power to fall away to zero at a particular point, which is not something that I don't think any of us have ever witnessed in the history of capitalism. It's so complicated and it's complex to decipher the different forces that are going on and that are at play.

16:14I reread Dan Juergen's The Prize, which is the 900-page history of the oil industry, a few months ago, and the first five or six chapters, which is the unseemly squabble around first Pennsylvania and then oil fields in Baku and Indonesia, the arrival of Kalust Gulbenkian. The characters, their machinations, the way they behaved, reminded me quite a lot of an industry I know much better, which is the technology industry and these big companies. So I think there's one lens where we can say this is really about power. It's about power seeking and people telling the stories they need to tell in order to justify their power seeking behavior, whether that story is, I'm doing this for the greater good of humanity, which is why I'm not paying attention to you because I know, or whether that story is, unless we do this, our adversary across the Pacific Ocean will get there first.

17:10There'll be curtains for us. But these are stories told by powerful men, and they're largely men, in order to justify less oversight and more support. And then I think the question is, what actually happens after all of that? Robert, you asked this question about AGI and the singularity. And what I would say is that the AI community, the scientists are split about what the meaning of that question is. So for the people who say, oh, we're two to five years away from this thing called AGI, where a computer will be able to do everything a human could possibly do and better, and then improve itself at computer speed, is disputed by people like Jan LeCun, who is a Turing Award winner.

17:54He's a senior AI scientist at Meta or Facebook, who says, well, AI today is stupider than a house cat, and we need to have some scientific breakthroughs to be able to get to that point of AGI. And there are other scientists who will say, a woman called Melanie Mitchell, who's associated with the Santa Fe Institute, who will say, the way that we actually think of intelligence is sort of misconceived. And this idea that you can define an AGI or even And an AI is a really un-nuanced way of looking at the problem. And my own take of this is that this is a really powerful technology that will have both industry-wide consequences, but also consequences in the way nations get run and their politics and their geopolitics.

18:42Whether or not the AGI boosters, the people who are more sceptical and think there's a different path, or the people who think it's entirely the wrong question to ask are right or not. If I had to put my money on it, I would say I sit with people who think this technology will continue to get much more powerful, but that there won't be a sudden dawn of AGI and we'll feel we're living in an AGI world. But we have to hold these different complexities of personalities and power of disappointment in the face of something that's still quite remarkable in our head all at the same time. And it's hard.

19:14It's hard for me and I do this full time. There's so much to cover in the tech world, isn't there? So let's have a quick break and then get back to it.

19:46There is one other, it seems to me to be, pressing public policy issue, which is implicit in everything we've discussed, but is just not on enough agendas of those in power. I mean, look, there are lots of service jobs done often from call centers. Somebody who is 40, 50 years old does have more experience and tends to be rewarded with higher pay because of their seniority. Already with AI, you can make an 18-year-old as efficient, as productive as a 50-year-old. And that means that wages are going to be compressed, that companies are simply not going to pay somebody who is older more. And that's a really significant economic phenomenon.

20:43And it's a phenomenon that has incredibly wide ramifications. about how you help those older people possibly to acquire new skills, whether you have to change. In my view, you absolutely do. This is, again, something we've talked a lot about on the podcast. I think AI means we have to redesign our welfare system to help people who are going to be put out of work acquire new skills or, arguably, in a world of wage compression, in a world where the productivity gains are being reaped by owners. Frankly, a welfare system that distributes some of that gain back to the people whose basic wages are being depressed.

21:24I am staggered by how little airplay any of this gets from our leading politicians, either in the UK or the US or anywhere in Europe. What do you think about this whole issue of the welfare system and AI? I mean, I think what you've described is quite likely to be an outcome, but companies will take this technology and they will see it first and foremost as something they can use for cost savings rather than to do new things. And some companies have not done that historically. Incidentally, I mean, Accenture, which is a huge services company, had to deal with automation about a decade ago and they re-skilled and up-skilled over 100 ,000 employees into new jobs.

22:02But they could do that because they were in a growing market. And I think a lot of bosses will not have those beneficial market conditions and they won't have the imagination, they won't be able to take the risk. So the path that you described, Robert, I think is one of the plausible scenarios, which is of that kind of wage compression. With the welfare system then, I think we need to understand the importance of the safety net. I mean, in the UK, we've unfortunately got the experience of what happens with a really dramatic change to local economic conditions without a good safety net. That was the debilitation of mining communities in the mid-1980s.

22:34And we can try to learn from that. And I think the specific things that you need to do are, first of all, you need to be able to talk and engage with the fact that these types of technologies, these types of managerial process changes are going to be happening so that the discussion can happen in a little bit more of an upfront way. And there can be some attempts to reduce the social stigma and the personal stigma for being in the wrong side of that change as it happens. And I think the second thing that we need to be able to do is we need to have an engine that supports the creation of new jobs in local economies.

23:13And that is essentially engines that support entrepreneurialism and business creation in all types of ways. It's already far too hard for a small business to get a 500 ,000 or million pound loan to expand their factory. I mean, that access to capital becomes really, really essential. And then you also need to have mechanisms for re-skilling and re-training people. The part of the gift of AI, if it's taken away some parts of people's lives, is that actually AI can be a very good re-skilling companion alongside human tutors right now. But those would just be sort of a patchwork of some of the things I think you'd need to do.

23:48I see where I live up in the Northeast is in the realities of when an industry declines, how it can massively impact an area and then impact it for decades later, you know, leading to really bad deprivation and people who feel like left and feel like they've got no place to go, not the right skills. And, you know, again, I've talked a lot on this podcast about how it would be better if we had lifelong learning, if, you know, people could actually not just think about doing education until 16 or 18 or 22 or whatever, they could actually just dip in and out all the time, because we've already seen how much the jobs market has changed now with the gig economy, it's not one job for life and all of that.

24:32And also the thing that I find that I bash my head against a lot is when I talk to, for example, further education colleges about this, because I'm very big into vocational training, is why we're not using technology and AI more in the education setting to allow it to be more accessible. And for example, you know, we know there aren't enough math teachers or science teachers. Why don't we mobilize and use AI and tech more to actually create our own artificially generated teachers and I know loads of teachers are going to kick off going just trying to get rid of us but to help not to hinder you know like virtual assistants in the classroom with you or you know if you've got one particularly brilliant maths teacher why can't that person work at seven different colleges virtually or we're so siloed into thinking it has to be you have to go into a building that's where you're going to learn and I was talking to the Chancellor of Newcastle University he was saying that actually their lecture theatres are empty most of the time now because everyone just watches the lectures from home or you know they do a lot of their learning online now and that's a big barrier the kind of education we need to think differently don't we about how we educate people to be able to live and work in this world we do and I think Steph you've talked about essentially ways in which we can maximize the talent, the education talent, and also reduce the barriers to access for people to receive that education.

25:59But they still need to have the motivation, the social cues, the economic support to be able to spend the time to do that, and then to get a sense that they can get a job after that. Because a lot of what you've described is, in a sense, it's already available. I mean, if you go to YouTube, you go to Coursera, you will find courses on everything from the basics of digital marketing to kind of complex financial trading and everything in between. So in some sense, it's already been accessible. And I think that there's a more systemic question, which is about a sense of self-belief. It's about a sense of where does this actually go?

26:40A sense of even if I have this skill, what do I do? And that's one of the reasons I talk about local businesses being the thing that's been important in this technology revolution. I mean, governments, and we're only a few weeks into the Starmer government, so effectively I'm talking about the previous set of conservative governments, tended to talk about the technology revolution and innovation from the perspective of the massive companies. They were really happy Google's got a new campus in London, and were really happy that Facebook, as it then was, is opening up a big office in central London.

27:10They didn't talk about the technology revolution as enabling small businesses all around the country to start to grow, to flourish and to employ locally and to actually understand their communities, which people who live in communities are the only ones who actually do. And so I think that's where we also need to give a lot of thought. I mean, effectively, what we're doing both with AI and other computing technologies is we're reducing the cost of being able to start a business. And so we need to make sure that people all around the country are able to do that. On that point I was talking to someone recently about how many people are learning makeup and hair and things online now and then going into the profession completely missing the whole education side of it they're just learning online not necessarily getting any qualifications in it but then being able to do it just as well as the people who have gone to college and done the qualification.

28:06And if they're earning and if they're happy and their customers are happy, that seems super reasonable, right? That's a great outcome. To take us back to what you were talking about earlier, Azeem, which I passionately agree with, one of the really striking characteristics of the riots recently is they were most acute in those parts of the country that suffered most in the deindustrialization of the 1980s, whether it's the closure of the coal mines or the closure of factories, the relocation of factories to China. We saw the anger of local communities manifested on the streets in, in many cases, really appalling criminal behavior.

28:56And some of that, some of that anger still stem from the fact that those communities were let down in the 1980s. And the thing that I am absolutely obsessed with is if we are correct that this industrial revolution is as important as any we've lived through, probably I think my argument would be more important than any we've lived through, we can't allow it to hollow out communities again. You know, we've got to look at the opportunities and then make sure that they are used precisely in the way that you talk about to see the creation of high productive businesses, particularly in those areas of deprivation.

29:39I mean, I think that's an absolute imperative in our public policy over the next year. So I think what the technology does, Robert, is that it enables that in a way that other technologies might not have done. Because you have these new cognitive assistants to your hand, because you have digital marketing channels, you can get started with a business, a smaller team initially, and lower investment capital than you might otherwise have needed. But you still need lots of other things to get going. And I think access to small business finance becomes really, really critical. And it's not the first£5 ,000 or the£50 ,000.