Azeem Azhar: Exponential

10 May 2022

GINA: Welcome to the Minderoo Centre for Technology and Democracy and our latest event, Azeem Azhar: Exponential. I am Gina Neff, the Director of the Minderoo Centre and we are delighted to have you here in person in Trinity Hall Cambridge and our online audience. Before we start the event, there are just a few housekeeping points here for our in‑person audience. Emergency evacuation, we are not planning on any this evening. If the alarm bell sounds, all attendees should vacate the room immediately use the stairs either the internal spiral stair case or the external stair case from Cherry Tree Court and assemble on the lawn. Smoking including e‑cigarettes is permitted outside the terrace room in the Cherry Tree Court. Exams, we kindly ask delegates keep noise to a minimum and don't wander around the ground as our students are currently revising or taking exams. I want to let you know that the event is being professionally live human captioned. If you would like to have captions select it on the toolbar at the bottom of the Zoom screen. Additionally there is a Streamtext captioning available. This is a fully adjustable live transcription of the event in your browser. No reason why that's not available to people in the room able to log on in the room. If you want to open this, the link is shared in the Streamtext. A transcript will be made available online after the event. The event will be recorded by Zoom and streamed to an audience on the platform by attending the event you are giving your consent to being included in the recording. The recording will be available on the websites shortly after our event this evening. For those joining online, questions can be asked through the chat function and moderators will share the questions during our discussion. For those of you in the room, when we get to this point in the evening, please raise your hand during the discussion and I will listen to your question, so please project and then I will repeat them so our online audience can hear them.

So, with that bit of housekeeping out of the way, I am pleased to be joined tonight by Azeem Azhar. Before I introduce him I want to say a few words of gratitude for his joining us this evening. Azeem has a remarkable ability to create communities. I was introduced to Azeem when someone in university communication said we want you to do a podcast, and those of you who know me know I am very late to the podcast game. I only discovered podcasts in 2023! I really am out of it when it comes to podcasts. So when they said we want you to do this podcast, I thought this is going to be terrible, waste of time, this is somebody who doesn't know what they're doing. They said no, he's a former FT journalist, he's really good. Like, trust us. Okay. When was the last time you said to a university communications team, they said trust you and you trusted them? It worked out. So I get to this call and lo and behold we're talking what I found to be an incredible conversation, that ends up being about self‑tracking, so on the other side is this wonderful polymath who can talk fluently social science and technology and finance and business and a delightful conversation. And two weeks later I am at an event and someone says you sound really familiar, oh, right, I heard you on Azeem's podcast and that's when I realised that Azeem Azhar has created an exceptional online community and have been witness to that through the pandemic days seeing how his exponential community has come together to support innovation, brainstorming, sharing.

So now, the formal introduction. Besides being a great all-round human, Azeem Azhar is an entrepreneur, investor and author. He is the founder of Exponential View the leading newsletter and podcast. Exponential View reaches more than 200,000 readers around the world. He is also an active start‑up investor with investments in AI, work from home and climate change. He is on the board of the Ada Lovelace Institute and sits on the World Economic Forum's Global Futures Council on Digital Economy and Society. Previously he founded PeerIndex, a big data analytics firm acquired in 2015. He has had senior roles in global media businesses and started his career as a journalist where he was a writer for The Economist and The Guardian. In 2020 he co‑authored The State of Climate Tech, a global analysis of venture capital trends in climate change investing with PwC. There's more. Tonight we are going to talk about his first book Exponential which explores the transition to the exponential age and was published in autumn 2021, a difficult time to launch a book. We are very, very lucky to be able to have a live in‑person event here tonight. I will now hand over to Azeem.

[Applause]

Azeem. Thank you, Gina, for that amazing introduction, I think it was probably written by my mum! It's really, really wonderful to be here. It is wonderful to be out and about, it's wonderful to have some sunshine as well. As Gina said the book came out in the fall of 2021, autumn and it's actually the best time possible to write a book because my plan in October 2019 was to spend December 2019 to May 2020 locked in my study, and it all went to plan in the end. So I was one of the few people who didn't get disrupted by the change in the work practices, as it were. What I would like to do today is talk for about 25 minutes and then move into a Q&A, just run through some of the ideas in the book, which is here. I picked up the last copy I am afraid, so I will be giving this out as a prize later because I left my copies at home. And the book has had quite a good reception. Generally quite well critically reviewed but I have two beautiful reviews from Amazon which I love leading out. Left wing slant, it is false advertising to sell a book without revealing the political bias of the author, unless you love the political left do not purchase this. On it goes.

The other one was it could have just been a couple of blogs but stretching into a book seems a bad idea. Substance is light. That is the beauty of the internet, which is this wonderful canvass for all sorts of opinions. My editor particularly will disagree with this because when I submitted the first draft there were 1,300 citations in it and he said that, you can't put a book like this with 1,300 citations so cut it down to 450. The heart of the argument is that we are starting with a couple of important axioms which I know for the socio-political economists amongst you will be very familiar to you from your own work, which is that we live in a world of complex socio technical systems, that is the political institutions, where our communities work, the way technology networks work, are all rather interrelated, and that they don't follow simple linearities, they are not a couple of yards of hearing an apple falling from a tree and hitting the ground. There are lots of feedback loops and lags and they are complex to analyse. The second is that, this is really important, and I dispense with this in the introduction of the book and I don't really return to it, which is that technology is deterministic. There has been not amongst the researchers and academics, but amongst the builders of technology a belief really starting in the late 19th century but accelerated in the mid‑20th century that technology was deterministic, if you didn't leap on the train I would be left behind and that in itself a lifetime study to unpick that. But I started a given that technology isn't deterministic, it is something we can shape and that we are shaped by. Which takes us really to the overview of the argument, some of which I will be able to get through when I am talking. It's that we are at a moment in history that certainly is quite unique, where we have general purpose technologies that are improving at really, really dramatic and radical rates. I call them exponential general purpose technologies. General purpose technologies are rare, they come along very infrequently. They are technologies that can be used in any part of our economies, in any part of a firm and have quite often quite long‑term second, third and fourth order effects. A good example might be something like the steam engine or electricity, where the changes that were marked by electricity were not simply in one industry. They are really in the reconfiguration of how work operated, what cities looked like, the availability of labour at home because we started to be freed of the cycle of dusk to dawn, the other way round, dawn to dusk when we get to work, unless we are vampires! And so a general purpose technology has these really dramatic shifts.

I argue that for the first time really since the turn of the 20th century, we have a series of general purpose technologies that are coming to fruition. They create all sorts of new potentials, as surely as the internal combustion engine put into the car and ship and plane, trans formed industry in society, these technologies will transform industry and society as well. But the second is that the world in which we live in is governed by a set of institutions. Some of these institutions are very, very formal and we need to fear them, like speed cameras and having to do our tax return every year, and some institutions are much less formal, they are habits and norms, customs. But institutions being sociological phenomenon, phenomenons of groups of people coming together will adapt more slowly and what that creates is this primary lens, which is the idea of the exponential gap, that on the one hand we've got technologies that are improving dramatically, that will have impact outside their technological domain and on the other hand the world in which we live is adjusts at this linear rate. For me, the exponential gap captured that sense that we have been feeling, I would say since if you are American probably since November 2016, and if you are British since June 2016, that, can I use the word crap, got to talk crap, robots are going to take job, Uber drivers are managed within an inch of their life, cyber-crime is going crazy, where does this all come from. I see it as that gap. My conclusion is that we need this systemic overhaul.

If we talk about the general purpose technologies, and I think it's really important to start with the technologies and understanding about how they arise and why they change the way they do, the four general purpose technologies that I see as being dramatically important are, the first is the one we are most familiar with, which is the fields of computing and AI. What those technologies are doing is they are essentially making it increasingly cheap to perform calculations. As a way of understanding this, this is an apple watch that I apologise to anyone who studies the dangers of wearing fitness trackers, for example, but this apple watch was given to me for free when I got a new mobile phone a couple of years ago, which is how common they have now become. And yet it's more powerful than the most powerful super computers that governments used to use to model the weather only back 35 years ago. It's now free. So you have this one technology which is computing and AI. Many of us are familiar with it because of the idea of moors law, but there are three other areas where there are GPTs emerging.

A second is in the domain of biology, because biology over the past 25 or 30 years, because of breakthroughs in certain theories, has become much more of an engineered discipline, and we are now able to look at biology less as a slow moving rather empirically driven experimental process but more as something that looks like a high through put industrial system and the technologies that have been developed in there that relate to gene sequencing, we are all familiar with that, if we have had PCR tests, but also to gene editing, protein reengineering and cell engineering, are creating a whole set of new industrial processes. The third set of technologies are those that relate to new energies. So as distinct to the energy technologies of the last industrial era, which was all these extracted fossil fuels, what we had was a set of technologies particularly in the shape of solar, photo voltaic, which are becoming dramatically cheaper on an ongoing basis and have different industrial configurations that have these second and third order effects. And the final most emerging technology is the technology of additive manufacturing. This is so emerging I have struggled to find a good emoji for it, which is why we have the factory, I have been struggling for two years on this. But additive manufacturing is the realm of digitising production, it's about being able to 3D print products, devices, components. A couple of years ago it was too expensive to do that at any scale but with the ongoing improvement in technology we are getting to a point where it is cost competitive to 3D print useful components and that again has really important economic ramifications. What is it about this idea of exponential though? It's important to bring that forward. I talked about my apple watch and how it was emblematic of the price of computing coming down. Because that's what this idea of moor as law, which says computers get more powerful every couple of years is essentially saying computing gets cheaper. But what I noticed in the research that I had done, and some others, this is a friend of mine, who puts this compilation together, that we were seeing these dramatic price declines across a whole range of different technology families, many of which we talked about in the previous slide. So, for example, the cost per watt of generating electricity through solar photovoltaics had gone from 5 to 20 cents. The cost of human genome back in the year 2000 when Bill Clinton was President in a more innocent time was about $1 billion, to sequence a human genome, today the price is about 100, 200 bucks. The cost to store the data has also collapsed as anyone knows, I remember the very first USB key I was given was an 8 megabyte key and now I get given keys with one terabyte, thousands of times more data on them. What does it mean when prices collapse? Because we may not feel it because our iPhone has always cost us £600 regardless. But what it means is that within the economy these capabilities get used and taken up much, much more and then new complimentary businesses emerge around them. If you think back to genome sequencing at a billion dollars a genome only a handful of people will get their genome sequenced and get that insight about the things that might drive their health conditions. At $100 a genome, pretty much everyone could get their genome sequence. A very, very large number of people as that price approaches a dollar many more people can have their genome sequence. That means when you go to see your doctor there's no reason they can't use that data when they come to think about your treatments. That is the important dynamic we need to think about when we look at curves like this. It is about the expansion of possibility into the economy much the other side of that is we don't necessarily know what complimentary products and services will emerge. And so who would have known, go back to 1950, in a time machine and ask someone and someone asks what is the most surprising thing about the time which you live today in 2020, and if you said to them I have this device that is more powerful than all the computers on the planet and can access all of the knowledge of humanity, and what I do is I use it to watch videos ever cats falling into buckets of paint, that is where we up going once these capabilities get as cheap as they do. Perhaps I can put the sense of change just visually. So about six or seven years a new AI technique came on the stage. Stunning, stunning technique, unbelievable, it was getting computers to draw images.

So here we are state‑of‑the‑art, 2016, the text prompt here is generated images of a picture, that is a baseball bowler, throwing a ball, and this is the best this thing could do, and it was mind boggling. I gave lots of talks about this back then. Last week, AI system given the text prompt, draw me a picture of Mount Everest as a cake in the style of digital art, and there we go. And that's the difference of five years of progress. Because one could also say, actually where is the raw data from which you learn. We often hear machine learning is about learning from data that already exists, where there is no Mount Everest of cake. So let's try to understand what drives exponential technologies and I really put it down to three fundamental drivers, that the most important of which is the first, which is learning. But also standardisation and networks are important recent drivers for this ability for technologies that improve at 10, 20, 30, 40, 50% per annum for decades which those technologists I have discussed have been doing. So if we start with learning, the key insight here pre‑dates the computer, the modern digital computer industry going back to 1936 and a man called Theodore Wright, who was looking at how engineers would construct aircraft. And he identified there was this relationship and I'm going to perhaps bring it home to everyone. Everyone here went into lockdown, obviously, unless there are some real rule breakers, did anyone start to bake lockdown bread? Any hands up? Okay, yes, I know you did, yes, a couple of people. How was your first loaf of lockdown bread? Quite flat. Did you bake any more loaves?

>> Lots.

AZEEM: Get any better?

>> Much.

AZEEM: Wright's law is essentially what happens, the first loaf of lockdown bread you produce is terrible and you waste loads of flour, by the time you do it eight times you are much, much better and more efficient. That is effectively that curve, it's called the learning curve. Engineered complex earning neared products often have learning effects. And these learning effects essentially say that for every doubling of cumulative capacity or doubling of cumulative out production of a product, there will be a learning that means the product is made more and more cheaply and for certain types of products, like semi‑conductors, silicon chips, that learning rate is extremely high. And that drives the cost declines that we see across many of these industries.

The reason I love that by the way, because as you know I have this left wing political bias according to that Amazon review, but I do have a bias because I am a human and I was producing something and that is what humans do, is that learning is such an important human characteristic and it's such a social characteristic, because we learn together, individually, in conversation, it's why we are gathered here today in Trinity Hall. And it is the thing that sets the clock speed for the improvements in the exponential age, this learning rate. And we have, we are getting better and better in many ways at walking down this learning curve. But there are a couple of other drivers of this exponential age. The second is the idea of standardisation.

So if you think back when you are at home and the batteries in the remote control are run out, as they often do in our house, what you will do, you don't have other batteries at home anywhere else, so you go to something you don't use often and rip the batteries out and drop them into the remote control. The reason you can do at that is because of standardisation. Standardisation enables larger markets for products which creates better incentives for people to produce them which reduces the cost. But it also means that learnings in one market transfer across to another market. And the hallmarks of industry in the last 60 or 70 years has been a dramatic increase in the number of standards that exist, that allow systems to interoperate.

When I first started work back in the early 90s I used one of the early speech synthesis applications and, speech recognition, and it was, by modern standards, it was not very good. But they had to write, build their own microphone, and have their own software and have their own programme, word processing programme because they couldn't plug into word perfect or Microsoft word. We have a great speech recognition in the room, but the point was that you had to build your entire thing.

Today, because ever standardisation, because of these things called APIs that live in software, the guys who want to build, the team that wants to build speech recognition only needs to excel at speech recognition, they can rely on standardisation to deliver everything else they want. The impact of standardisation is that it can accelerate the speed with which products can come to market, which drives the exponentiality. One of my favourite things of standardisation has been in the space industry. A decade ago a couple of academics from Berkeley University came up with the nano sat, they said if we make them out of these common Lego bricks you can plug together we can reduce the cost of building a satellite. Because those satellites have the same form factor, it will mean that the launch vehicles can be simpler and the cost of launches can come down. You can go on to a website nano sat.eu and look at the number of satellites launched year by year and there's this curve this ticks up in 2013 and goes crazy after that because standardisation entered that industry. So standardisation is really important. Then the third thing that's important is networks. There are three that matter, finance, trade and that of information. I will go into more detail in the book, but the key thing about networks is on the one happened they make markets much bigger, which creates bigger opportunities.

But on the second hand they make information knowledge and learning travel much faster. So one of my favourites stories about this is a girl called Laura Connell, who is probably about 20 now, 17 when I met her at a science competition and what she had done as part of her second last year at high school, entered this science competition was built a working machine vision model that could look at scans of PAP smears and determine more accurately than a typical doctor whether they were abnormal or not. There had been a cervical cancer scandal in Ireland earlier and so lots of high schools entered this. The fascinating thing was that Laura, who was not even from Dublin or Cork, from a really small town in Ireland, had never programmed a computer before and her first programme was this thing, that out performed a typical consultant, in a lab scale, not actually in a hospital. But the technologies she used had only been written up three years earlier at Stanford University. And actually hadn't gone through the peer review process to get into a journal at the time because the internet through its networks had compressed the time it took to get that knowledge into the market. So those for me were the three drivers this idea of the exponential age. I felt as someone who has been working in the tech industry for nearly 30 years, that people are very used to mostly men standing on stages, very breathlessly, saying "and one more thing, this is going to change the world, this is our greatest ever and this will change the way you live."

We can remember the launch of the Segway that was going to change personal transportation and a million other things. But what we need to do is make sure that we don't get too sceptical and so I went to a lot of trouble to try to explain why I think these processes happen and to demonstrate in the first few chapters of the book that they actually do indeed happen. So if I have persuaded, maybe I have, maybe I haven't. That gets us to the simple visualisation really, which is that you have these exponential technologies and they through a process of combining and recombining and being taken on by organisations and by firms and by investors, become real, and they tend to accelerate in this way. This curve, it's a mathematical curve b you can overlay it on top of the number of iPhone sales. You can overlay it on top of Netflix's revenues until this year by the way. You can overlay it time and again on the rate, current rate at which Norwegians are buying electric vehicles and it maps pretty well.

What we have of course are these institutions, norms and expectations that are changing at this much, much more sedate linear pace. It's practically flat if you ever watch the American Congress question Mark Zuckerberg and the other internet bosses. And I think the challenge that we have is to address the question about what happens in this space. And there are two ways to think about this. What can you do about this angle, this gradient and this rate of change? And what could you do about this? Is it a case of pushing this up, is it a case of pulling this down? And in a way that's what I try to address in the second half of the book.

So this is such a busy slide and I'm really sorry, I was told not to do busy slides and I said there only be one, I want to explain how this all comes about. So at the bottom, layer, the lowest layer you have the core technologies and below these technologies there are component technologies which are in many cases because of learning, because of standardisation, because of networks, improving really, really rapidly. They can then be combined to build applications like AI and drones, which are also really improving dramatically. Think back to the baseball picture to the cake, mount Everest made of cake which sounds delicious and they then also find their way actually into the real world. Firms and business models. You can look at a company like Facebook, which frankly is probably not the most loved company in the world, now called meta platforms and yet despite coming under that degree of scrutiny, still grows and still elects to find new avenues in which to grow.

These firms are on that trajectory. You also see it within how these things get deployed frankly amongst state actors, in particular if we think about how the use of cyber-crime and disinformation has emanated from various state actors. But the second half of my argument then relates to what this exponential gap is and the impact on our norms and our informal practise, our institutional arrangements. Perhaps one example of that informal practice is this is a very tangible example of the exponential gap, which is one that I live with nearly every day. Should teenagers be allowed to use their smart phones at dinner? When I grew up, no‑one used their phone at dinner because the phone was in the corridor and if someone called you during dinner time, someone would be despatched to say we're having dinner and hang up. Of course, phones have found their way suddenly into our homes, many times over, and we still don't know what the rule ought to be. And of course, as we know as the researchers you are, our first assessment about whether these things are actually good for us or bad for us is often no more than reading the tea leaves, because it's not research, it's not academic, it's not been demonstrated. We still don't know. There's a tangible version. Let's touch on my argument that this is systemic.

I split this out into five big categories, the naivety of a first time author. I look at markets and the economy and the trends that I see emerging in the exponential age are network effects, the growth of platforms, dependent on intangibles, they are dependent on software on patterns on creativity, on ideas, the things that brains bring to the table. They are not dependent on closing off markets or extracting things from the ground, or fashioning bits of iron into pins sharper and sharper. In the area of labour the trends I see relate extensively towards the balance of power between these powerful firms and workers specifically around how work gets managed and how work gets parcelled out.

As a brief spoiler I'm not a big believer in the idea that robots take jobs. Within the space of economic geography, the thing that I see and argued is that the trends of these technologies even though these are trends of largely centralisation and accretion of power to central increasingly powerful firms, that the trends here of the technology force us towards a localization that is perhaps sits accountable to the globalisation we have seen over the last 50 years. In the area of conflict, what I see is these technologies are very dual use and not necessarily technologies of peace, they are actually more likely to be technologies of configuration. In the area of politics that we have all lived through and felt the frustration of the social Ned works and social networks, the notion that our space for public discussion, but also other parts of our lives that we have kept out of the realm of the market. For example, the relationships we have with our friends. Are being taken over by privatised rule makers, that technology firms in particular have found their way into these personal interactions, or into a space where we have felt that perhaps market rules shouldn't be the thing that determined behaviour. I think about that in terms of as the privatisation of the public's. And the challenge with looking at all of this and why I tried to get it in one book and on one slide, is that I think this requires a much, much broader conversation than a point by point conversation. That there are really close relationships between each of these areas. One of the themes that I think runs through all of this but I didn't use this in the book because I didn't want to come across as someone who was super, super left wing, which I obviously failed to do, is the current theme is really about changing the power. Shifts in the power that emerged here. Who is it going from, who is it going to? It's not necessarily going from the powerless to the powerful though there are trends like that. It's sometimes going from one powerful group to another powerful group. But that is one common theme that underlies it. So there's a lot in the book. I know that, but I hope I have given you a good summary so far. Thank you.

[Applause]

GINA: As you are getting the questions together, and the online audience is getting their questions together, I've got a few questions of my own. First Azeem, thank you. That was great. It's so great to have you here. I would love to pick up where you left us. Because in the book you talk about the new world disorder, and this gap between where our technologies are and how they're growing and how our institutions are not growing. That leads us to all kinds of problems. As a sociologist this is what I am very interested in. I think writing there in 2019, 2020, you seem now prophetic in talking about the new wars. We find ourselves in a moment where there's a war in Europe and it is a very different kind of war than we have seen in Europe before.

My question really is do you see that disorder that's coming from the exponential gap, do you see this as permanent, or do you see this as a result of the gap? Can we ever catch up in our institutions?

AZEEM: No, it's such a great question. I've got a mic and I would say in that chapter I do have a big section on the Turkish drone which if anyone has been following the conflict in Europe, is a highly sung, there's a song about it about the Ukrainian fight back against the invasion. So I think that we do, part of the challenge in the new world disorder, we have run through those ideas, was that the idea that any conflict is a balance between the cost to the aggressor and the benefits that they might see.

If you make the costs extremely high, the best example is mutually assured destruction, you stop aggression. If the costs get lower and lower because you don't have to send soldiers to die, you can send a drone, or you can just constantly needle someone through cyber fraud and cyber-crime, you construct this moment where essentially you have a persistent set of aggressions and that's where that idea of new disorder comes about. Then the question is what's going on of course is that drone technology is making physical attacks cheaper and cyber-crime is making it cheaper but also we are much more vulnerable. Because every time, I was never, until I was, until 2005, no‑one would ever consider little old Azeem is a national entry point, if you wanted to invade British infrastructure. You would have to send ships and tanks somehow to us. But the moment I was connected permanently to the internet I was an entry point to the British network. If you have bought a smart light bulb, of you have created a weakness for an adversary to get to you.

The question is what would it take for those sorts of things to be addressed. I think we do have an adjustment period rather than saying it's hopeless. The adjustment period comes from things like the IOT devices, it comes from saying instead of internet, devices being completely unregulated, anyone can sell them and no‑one has to worry about software updates, we might start to say, well, if you are going to sell them, you can earn this badge of approval that says we'll provide you with security updates. That's an institutional change that we can grab hold of. But then the same comes to our institutions that we use to govern these types of issues, what's been really interesting with the Russia Ukraine things and some things I couldn't have predicted is that the UN has not been very useful. But other institutions have been quite useful. The EU being one. But ones that we perhaps are not as obvious to us like the Five Eyes, which is an international group of five countries, US, Canada, UK, Australia New Zealand, and five Is could only exist in the exponential age because how else could you have spies in five different countries looking at data in real time, the same data. So there are new institutions that can emerge. But they do take time. So I think we are in a period of needing to make those adjustments rather than hopelessly lost.

GINA: Another area of hope or hopelessness might be the impact on our notion of citizenship. That sense of what is to be done is certainly mobilised and thought through how we experience the world in social media and the kind of new world, not only is it through our IOT devices that we can be brought in as ways that adversaries might reach us, but active disinformation campaigns, information operations and misinformation are ways to seed, to sow seeds of confusion and distrust and I'm not content. How do you frame the exponential age in terms of citizenship?

AZEEM: I think it's fascinating to see again what's happened in Ukraine because the Russians who are meant to be great at misinformation and misdirection have not been so good at that in the last two‑and‑a‑half months and the Ukrainians have won the mean war really just in remarkable ways. I think it demonstrates that perhaps these tools need to be used as part of countering some of the strategies. A great example that pre‑dated that invasion was in Taiwan, they do have someone who is in charge of Memes within the digital industry and they are doing a lot of meming because they realise that is the to communicate. It is a bit like saying Nixon was the first President to understand television. Who is the first President who understands memes and, Zelenskyy, so those are the new tools we end up needing? But there's another really interesting dynamic to this and before I get there I want to steer away the bit I don't really understand, I don't quite understand why the US is so different to other places. I don't quite understand the dynamics of how, say, science scepticism emerged around COVID and I'm reading it, I am reading some of your colleagues as well on that. But one thing I think is quite interesting about prospects for citizenship are where are the spaces for being active in public policy and what does the technology allow. I will give you an example. In power generation for the last 100 years power generation has been a very industrial process. Somewhere on the coast for a billion pounds is a gas power station that pumped electricity in one direction, it's regulated nationally and little you or me can have no real say and no sense of agency over that. And those decisions are taken very, very far away from us. But as we, some of these new technologies allow for very decentralised power generation. So in Cornwall for example there's a local electricity market which is lots of people and farmers sticking solar panels on their houses. And now running into the Government's problem which is that they are things they want to do with the assets they own in their area that are regulated by some national market. That's an exponential gap as well. It's not as sexy as saying democracy is going to die because of Facebook but it's a practical one and shows up in South Africa as well where cities like Cape Town are capable of generating loads of electricity but the national authority is sort of working with a national plan. What I think is interesting is this idea of the locus of democracy being able to emerge in new areas we never thought we could participate as citizens. I find that particularly empowering idea that together with my neighbours, what could be more democratic, what could be more participatory, or more citizen like than that. We are sitting there discussing the energy supply and storage and what happens during the night and so on. The technologies, because they don't cost a billion pounds, they cost two grand enable that.

GINA: I saw our first in‑room question. I need to repeat it for the online audience.

For those on the online audience, the question is: is it about dampening, is it about bringing up the level of institutions or dampening the growth on the curve on the exponential technologies, what are the actors and what is the role of the state and what is the role of capital and venture capital?

AZEEM: Really important question. One of the reasons I give these process underlying processes of the learning curve is to identify that there are these underlying processes that are quite hard to avoid. The problem with knowledge as Eve discovered is once it's out there it's out there and can't go back in. And so I think it's quite hard to slow those processes down. I thought quite hard about this question in the sense of could you as a given state say we're not going to allow these types of developments and the problem is twofold. The first is in a competitive market, there's a good reason why everyone in Russia wants an iPhone rather than the Russian equivalent of an iPhone and so you will be out‑competed in a way. There's a second reason, which is that these technologies are at some level tremendously beneficial. For example, the carbon intensity of the UK economy is much lower now than it was 25 years ago and will continue to be because of computation and renewables and because we don't know the answers today, we need to allow research and basic science to explore wherever it needs to get to.

But that said, I think even if it's hard to change the gradient or second derivative of that curve, we can actually shape that direction. So I think the weakness of showing it as a line is that actually it's not a line, it's also directed. And how does that direction get shaped. I think that direction gets shaped by values and it gets shaped by storytelling and narrative. And so we can participate in that. One of the big shifts, just before I wrote this book, I wrote a big report on venture capital in climate change, where the amount of venture capital that was moving towards climate change was increasing dramatically in the year since I wrote the report, proportionately it more than doubled but the real trigger point was Greta Thunberg creating a narrative that somehow just put enough, removed just enough friction of belief that the climate narrative changed. So the way I think we shape what's important is through our narratives, taking Francis Horgan and Christopher Wiley seriously, through starting to say actually Amazon workers should be allowed to unionise and by creating those stories. The core technology I think we generally benefit from and that to me is all about participation, and one message I realised after I wrote the book, is that in a way part of our democratic participation is actually caring about these things. In a sense for a good 15 or 20 year period we were willing to be happy with faster I phones rather than having the discussion that we're now having. In a way we are the custodians of this. Poach is going to poach and so we have to participate and ask those questions.

GINA: We have time maybe for one, possibly two more questions.

AZEEM: I will give shorter answers.

GINA: What are the drivers of the growth of institutions and what should we be doing about that?

AZEEM: Great question and I do tackle that institution question in chapter 3. I didn't quite put it into the slide. But I think an institution can be designed in particular ways.

One of the things we want from institutions is we want them to have some sense of purpose and some sense of perhaps rigidity is the word I am after, or transparency and consistency, otherwise they wouldn't be institutions, but I think there are things you can do within institutions that relate to the missions that they have and their own ability to recognise the importance of change and so one example there is some of the regulators in the last five or ten years have started to say we're regulating an area like banking, but there's a lot of innovation that is coming through that is really good for consumers, that the traditional banks can't do because of their own issues. So what we'll do is we'll create, change our operations and create sand boxes where new innovations can play while we learn about them. I think that is an institutional shift that changes the way the institution operates that allows them to look at what this innovation is and to start to bring it on board with a slightly better understanding of the risks. Another model and this is a Chinese model, is the model of saying the phrase was let the bullets fly, which is let experiments happen and if we see things we don't like, we can come down quickly, which they did about six months after I wrote the book. That may not work in our democratic environment.

One of the pleas I made, I talk a little bit about three design principles in the book that are flexibility, so starting with the assumption that things may change. And that is literally how do you pack when you go to a holiday to Amsterdam or Belgium verses when you go to a holiday in the Seychelles. You pack a range of different clothes, whereas in the Seychelles you take your swimmers.

The second is resilience, which is how do you design an institution that is going to be able to with stand uncertainty on the downside and we haven't generally for the last 40 years worried about resilience as a design parameter, we have worried about efficiency and the third and most important one is this idea, in fact in this wonderful book that you have got on your desks, sorry for the online audience, the second last, essay, I argue about commonality because a lot of the issues I raised were about the lack of participation and the lack of common goals in these arenas, and an institutional design that starts to think about collective based approaches to what the questions are I think secures the participation that you need to give it some sense of legitimacy and some sense of responsiveness.

One of the things, again I couldn't get this, this is a mainstream book so couldn't get into the detail about this, was that a lot of what we have to do is, have had to deliver through institutions is so deeply technical that it's really, really hard for a non‑expert to get into it. So that's why we have to ask them to be much more principles driven because non‑experts can understand principles even if we don't understand technology mechanisms, your question is a really brilliant one. Hopefully my second book will answer it in more detail. There are some principles that are in there.

GINA: That's a great set‑up for the next time we will invite you here. We are at the end of our time today. I want to thank Azeem so much for joining us. Thank all of you for joining us both in person here in Cambridge but also online. I would like to call your attention two of our upcoming events. Next week online May 17th we will host Lisa Parks. Lisa is a MacArthur Genius Award winner, she will provide an overview of her book in progress mixed signals, media infrastructures and globalisation and it will be a wonderful add‑on to tonight's event.

Then on Friday May 20th we are hosting in person only Francis Horgan to discuss tech and big tech's role in national security and that will be in conversation with me and Professor John Naughton at the Babbage lecture theatre. So please look at our website for further events. For those of you joining us in the online audience, we would appreciate if you would fill out your short feedback questionnaire after the end of the session and the link to that survey will be sent via event bright and that might come to you those of you here, it comes to you in person too, give us feedback, we would appreciate it. Look at our events in the future on our website, mctd.ac.uk, thank you all so much. Join me in thanking Azeem.