Koen Smeets: Welcome back to the interview-series on the socioeconomic consequences of disruptive technologies by Rethinking Economics NL. Today we will be focusing on the sociopolitical and economical environment of the Europe. We will especially be focusing on how this environment is influencing how these technologies are shaped and regulated. For that, three world-class experts have been so kind as to make time for us today.
Firstly with us today is Diane Coyle. She is Bennett Professor of Public Policy at the University of Cambridge and co-directs there the Bennett Institute for Public Policy. She's also, she was a Professor of Economics at the University of Manchester. She specialises in the economics of new technologies and globalisation, particularly measurement of the digital economy and competition in digital markets. Previously she was the Vice Chair of the BBC Trust, a member of the
Competition Commission, the Migration Advisory Committee and the Natural Capital Committee. She was the author of a number of books in economics, including "GDP: A Brief but Affectionate History" and "The Weightless World: Strategies for Managing the Digital Economy". Secondly with us today is Andrea Renda. He is an Italian social scientist, whose research lies at the crossroads between economics, law, technology and public policy. He is Senior Research Fellow and Head of Global Governance, Regulation, Innovation and the Digital Economy at the Centre for European Policy Studies. From September 2017,
he holds the Chair for Digital Innovation at the College of Europe in Bruges in Belgium. He is also a non-resident fellow at Duke University's Kenan Institute for Ethics. He is a member of the High-Level Expert Group on AI as well as a member of the International Advisory Board at European Parliament’s Science and Technology Options Assessment. Lastly with us today is Charlotte Stix. She is an technology policy expert with a specialization in AI governance. Her PhD research at the Eindhoven University of Technology critically examines ethical, governance and regulatory considerations around artificial intelligence. She is also a Fellow to the Leverhulme Centre for the Future of Intelligence at the University of Cambridge and an Expert to the World Economic Forum's Global Future Council on Neurotechnologies.
Most recently, Charlotte was the Coordinator of the European Commission's High-Level Expert Group on AI and she was also awarded as a 2020 Forbes' 30 under 30 of Europe and a Young Global Shaper by the World Economic Forum and in her spare time, Charlotte runs the bi-monthly EuropeanAI newsletter with >1900 subscribers. And with that, I would like to move to the first question for professor Coyle. Could you tell us more about the digital economy, especially its main divergences from the economic perspective from the “traditional”, non-digital economy? Diane Coyle: The first thing to say about digital technology is it's what economists call a "general purpose technology", which means it's got wide uses throughout the economy and it affects many different activities and sectors. And of course we've experienced that particularly since 2007,
with the arrival of 3g and smartphones and we now, particularly this year with a pandemic, understand how much digital has transformed our daily lives and we use it all the time. In terms of its economic characteristics, there are some differences. One is that, like some previous technologies, there are very high upfront costs and then very low marginal cost of using lots of digital technologies, and so there are these increasing returns to scale. It's been true of other industries in the past, but this is one of the important economic characteristics. Another is that there are what economies call network effects, which means that the more people use the technology the more you benefit from it yourself. This was true of telephony but it's also true of lots of digital applications. And together
these create quite a unique kind of market dynamic, winner-takes-all dynamics. In that, you've got to cover the upfront costs. So companies that enter the market are going to make losses for a long period and they need that funding, they need a large market to help them cover their initial costs. And they reach a critical mass at some point, when the market
tips towards them and so in many of digital markets you find one or two dominant players. And this is something that's obviously come to the fore recently with lots of countries and the European Union just this week, publishing new regulatory and competition approaches to digital markets, because I think we have steadily realized the importance of them in our lives and the fact that we are very dependent on these markets. And I think there was a great example of this dependence and therefore vulnerability with some of the hacking that occurred just recently of US government websites. And this has nothing to do with the underlying technology, it wasn't to do
with internet security, it was to do with economic concentration. So on the same day Google went down and the US government got got hacked, and the vulnerability was about the market concentration rather than anything to do with the technology. And so we're at a point now, where a lot of governments and authorities are starting to think very hard about what the implications of these technologies have been. On the one hand, like any big technological advance, you've got the potential for improvements in productivity in people's quality of life. And on the other hand, because of the special characteristics, quite distinctive vulnerabilities in our economies and societies as well. Koen Smeets: That's very, very fascinating,
I’m sure we'll move also very to the regulation aspect of this. But before we do that, I was wondering, could you tell us, we often see technology, it's often framed as something inevitable that just happens to us. But I’ve seen, heard you argue that it is actually a social, a construct by its social-political environment and therefore we can influence how it's shaped. Could you expand on that? Diane Coyle: Sure, let me give some examples. I talked about the need for digital companies to raise enough funding to cover their losses initially. This is one of the reasons why the giant digital companies are US or Chinese, they've got very large domestic markets that they can expand into, but also they've had the funding to cover those losses because the structure of their financial sectors. And that's been harder in Europe, so Europe doesn't have the same dominance in any of these markets. Another example
is about data. And there's a lot of debate now particularly about personal data and understanding the implications of big digital companies taking a lot of individuals and data about their activities and transactions, and using that to sell new services or to target advertising. And that's possible because there's been a presumption that data is something that can and should be owned, and it's owned by the companies that are collecting and structuring it.
And that's that's a presumption that has sort of come about by chance really, because of the structure of legal ownership, but that's something that could be legislated. We could legislate for an entirely different data economy that was not about ownership and transactions in the market, but about terms of access and who has permission to access and use different types of data, which is much more linked to Elinor Ostrom’s view about how you would organize the collective use and distribution of resources. If you think back to previous episodes of major technological changes, they've had very different outcomes. There's a lot of debate now in economics
and policy communities about whether automation will accelerate and whether it will lead to a lot of job-losses and inequality. If you look back at previous waves of technological change and compare and contrast them, the Industrial Revolution and the 1950s and 60s, they both saw a lot of automation, a lot of change in the labour market and the kinds of jobs that people people did, but the 50s and 60s saw increasing employment and reducing income inequality. And that was shaped by the social and economic context of the time, and the fact that it was imperative after the sacrifices populations had made during the Second World War, to ensure that those technologies delivered wide benefits for society.
Koen Smeets: So then seeing how this, the importance of such a social political environment, Ms Stix, could you tell us more about what this environment looks like within the European Union? Charlotte Stix: Sure, I mean I think for the European Union it's really important to also mention that, recently in the Digital Markets Act and the European Commission has actually called for fines for up to 10 per cent of the annual global turnover for online gatekeepers. So those are the really big tech giants, you know the ones that do have a lot of the data, that do manage a lot of the access and what can be done with our data. I think what's also important to notice, you know in terms of future technologies, emerging technologies. Europe has had quite a leading position with the GDPR in terms of how we think about of these things, and how internationally different actors look towards Europe and sort of build on the work that Europe has done in this field. You know, it is lacking sometimes in,
you know industrial, how to say that, in startups and SMEs, that is a criticism that is valid and that is being tackled, but Europe has really had leadership position in terms of ensuring that technology is used and deployed for the citizens and for the individual and empowering and enabling those individuals to really, you know look towards what happens with their data, what happens with the technologies they use and ensure that they are protected. I think there is also something important to mention, that oftentimes you have these people that would say that Europe is regulating itself into irrelevance, which I mean, it's, it is a valid criticism depending on where you come from. But with the recent white paper on AI, you see that Europe has really tried to set the bar very high again for consumer protection, for ensuring that individuals have access to so-called trustworthy, human-centric artificial intelligence, which is a key element in the future and hopefully puts Europe in the leadership position. Koen Smeets: And could you expand more on the environment, how they're tackling the weaknesses, through CLAIRE and ELLIS? Charlotte Stix: For sure, I mean CLAIRE is one of the sort of thriving ecosystems, networks, academic networks in Europe, and they have done a lot of work also proposing their lighthouse centre on artificial intelligence, which would be a sort of CERN for AI, where part of the idea is that Europe has great research institutions, that's a fact. Unfortunately our retention rate isn't as good as it could be, so you know having a really ambitious project, a sort of moonshot project, would potentially attract, not only researchers that have been trained in Europe and are European, but also those from other countries, and it would also enable and empower industry players to potentially, you know harness European researchers from that center. But you also have a couple of other networks and tools. So you have the public-private
partnerships, you have the coordinated plan on AI, which is sort-of at a member state level where various countries collaborate to counteract fragmentation in Europe. I mean, we have to remember that the European Union isn't one country, so to sort of ensure that really everything is pulled together and that there are strategies across countries borders. You have the Digital Innovation Hub networks, where again there is a really big effort and a push to ensure that the industry is working with governments, with academics and researchers, to think about how to increase the, to increase the access to this technology, to increase funding, but also to enable the public and other actors to really engage with these technologies in the future, and that's something that doesn't quite exist in this formal variation in other countries that are comparable.
Koen Smeets: And what I was quite curious about, is that Europe is often portrayed as far less of a world player in AI than the United States or China. But I do think that, if I for instance looked at you, what you've written, it does show that Europe is actually quite a player. Could you expand on Europe's position in AI and how do you expect this to develop? Charlotte Stix: Sure, I mean as I said before, I think it really depends on your angle, and where, how you weigh what you think is important. So if you weigh the number of startups and the sort of funding for startups as really important and key to AI-progress, then that might give a completely different result than if you weigh, as I’ve said before, efforts towards AI-regulation, standardisation, certification, and consumer protection.
Now, one of those things is quite sexy, the other one might not be as sexy. That doesn't mean that it isn't really important, doesn't mean that it's isn't vital, and doesn't mean that, it doesn't give other countries and players a sort of direction that is being followed internationally when you look at, for example what the United States is currently looking at. So it might not be the most ideal, in the shape and form that the regulation, or regulatory framework, is in right now, but it is still a guiding structure for other countries on what they actually want to do or not to do. And I think that, in and of itself, is a leadership position. It might not be the leadership position, but that really does depend on how, how you weigh what is important to you. Now, you could also say that the European Union is putting in a lot of effort on ensuring an ecosystem of excellence, which you know looks at technical capabilities, which looks at funding for artificial intelligence, because of course if you regulate, you need an equal technological, you know pool to regulate, and it's important to enable that pool, and to really ensure that the industry is thriving as well. So I think Europe has a really good chance here, to have a leadership position in ethical, in trustworthy artificial intelligence, and which I’m sure that Andrea will expand on, and it is important to not undervalue that and underplay that, just because it is maybe not as sexy or cool, and I say this on purpose, than having really hotshot startups and really big industry players.
Koen Smeets: I think that's very, very interesting, and with that I actually also want to move toward Professor Renda. I was really curious, how exactly are we regulating these technologies, and how do you see this? Andrea Renda: Well, all digital technologies, of course, they are a big spectrum of solutions, and incorporating hardware, software and various different solutions, applied in different sectors. So there's no single regulatory framework, obviously in Europe that encompasses all of them. And indeed, for a long time the regulatory framework has been very light-handed in Europe,
but just as it has been in many other parts of the world. We have to remember that, when the internet started permeating our lives, like in the early 90s in particular, concrete, intentional choices were made not to regulate the internet, and to leave the baby grow-up, let's say, maybe spoil the baby a little bit, with the lack of regulation, in the belief that this would lead to the so-called "Permissionless Innovation" and a very ecumenical development, let's say, of the internet, where there will be some value for each and every player, each and every user, from this fantastic ecosystem, nurtured by the effects and the and the phenomena in economics as well that Diane was mentioning before. Network externalities, potential sequences of one generation monopolist. So high market contestability, very turbulent, but also very dynamic and creative environment. Now obviously, we've seen that things have not exactly gone in that direction, meaning that after a first, if you wish, very creative season in which, being a one-generation monopolist, did not necessarily mean having a guarantee of remaining, the one that rides the wave in the subsequent generation. It's a sort of early big bang, if you wish the dynamic environment. The situation is very
much crystallized, along centripetal forces that have generated four, five, six large players that accumulate and capture most of the value that is being generated on the internet. And this is something that happens in unregulated environments in most cases, because the overabundance of the information on the internet is also, and the modularity of the products, that are largely consumed and distributed over the internet, has generated also, a, what if you want to go back to the early days of economics and behavioral science, for example, what Herbert Simon would have defined a situation in which a wealth of information creates a poverty of attention, and a situation in which a few players that have captured the attention of the end users, can monetize their attention in various ways. So this we know where this is led. In many respects, inequality in many markets, so see what has happened in the United States this year [2020] in terms of the economic security of those jobs, right. From 300.000 people on
unemployment subsidies at the beginning of the pandemic, to 30 million three weeks after that. So it's, just to give you an idea of what, maybe some economists would would declare and would describe as being the marvelous, flexibility and of the of the US labour market, I’d rather would define it as a lack of economic security, and potentially over time a lack of social cohesion. And that is, also brings me to maybe a comment that is more general, with respect to the to the questions in the that you have formulated so far and the answers you've gotten, which I very much agree with of course, is the fact that, technology is a means, and that's as such it is not regulated at EU-level, meaning regulation tends to be technology-neutral, meaning what you regulate is the products and the services that use different types of technologies over time, with some exceptions that I’m about to spell-out that are more recent, where we start already building more technology-specific regulation. Technology is a mean, so we should treat it as a means. If we think that artificial intelligence is an end, and I come to my experience in the High-Level Expert Group on AI, on the first day the European commission asked me and the other 51 colleagues there, we want to become more competitive in AI, and let me provocative here a little bit, who cares, right? I care about meeting the Sustainable Development Goals. If I
can do this with AI, let there be AI. If I can do this without AI, well that's, then let's see what AI does. But if AI, if really aiming at becoming competitive in AI means, having a perhaps an exaggerated approach towards filling factories with robots, and without creating a meaningful complementarity between artificial intelligence and human beings, and maybe it means that we will not meet our Sustainable Development Goal aids of achieving full and decent employment for everybody, well that is I think a very distorted way of looking at public policy. So I’d rather have a different optic towards regulating technology where I think that my vision, well I base my regulation on the principles, which are, in the case of the EU, very much nested in the treaties, and in what we call the European Union values, although we would have, we would need to have another interview about what these are in particular. And we also are clears about, clear about the goals, the vision that we want to realise in the medium term, and then we deploy our technology policy in that respect. Okay.
Now, just let me wrap up on this, and I think in the High-Level Expert Group we have then taken at least partly that direction of trying to treat technology as a means. So not sacrificing our medium-term prosperity on the, and sustainable development, on the altar of digital technologies. And then over time, I think, to close the loop that I’ve opened in my answer, the largely unregulated environment that it started with, I don't know, the WIPO treaty in 1996, the Telecommunications Act, or the Communications Decency Act in the United States, the Information Society Directive, the E-commerce Directive at the European Union level in the early 2000s, 2000 and 2001, that loop is now being closed, because the European Union, maybe before the United States as realised, maybe through competition cases first, and then later through regulatory attempts, that the time was right to start rebalancing the power that has emerged through those centripetal forces that I was describing before. And then starting to experiment first with antitrust laws, and then realizing over time that maybe antitrust is not enough. And then starting to look into what member states have had for a long time, rules on superior bargaining power, rules on abuse of economic dependency, and bringing that and them back into the EU-level competition policy and related regulatory interventions, to build what today we call the Digital Services Act, the Digital Markets Act, but before then the Platform to Business Regulation. In all this, and sorry Koen, you already know that, I know you want to ask another question so I’ll stop in 30 seconds. In all this, the one off attempt that has been a sort of game-changer,
albeit and perfect in this process, is certainly the GDPR. So an attempt to introduce in this largely unregulated environment, with also social norms and self-regulated behaviours that are very different from what was happening in the real world, in this environment, introduction of a non-negotiable set of rules. Maybe not fully complied with, maybe not having the impact one was expecting at the very beginning, but still an assertive decision to make something non-negotiable, for example, or rigidly protected, which is personal data, has been a game changer in terms of turning the tide towards understanding, and this is something that is shared in the US, in the EU, and in most other parts of the world, think about Japan or other countries, that the internet has to be regulated and digital technology has to be regulated. Koen Smeets: I think it gives a beautiful overview on many topics including the environment and how this has been historically shaped in the European Union. Thank you for that,
and I was also very curious if you could shortly expand on what exactly have been the recommendations of the High-Level Expert Group on AI. Could you tell us more about the Ethics Guidelines for Trustworthy AI and the Investment Recommendations for Trustworthy AI? Andrea Renda: Yeah, so the High-Level Expert Group was given two mandates basically, two products to develop. One was this Ethics Guidelines and on Trustworthy - well Ethics Guidelines on Artificial Intelligence actually, the original mandate. And that has been I think, the most fruitful elaboration of the High-Level Expert Group because we did two things, cutting a very long story short. A subset of the group, in particular the academics in the group, have given a precise direction to the work of the High-Level Expert Group, without replicating the dozens of dozens of ethical principles that are already available, from bioethics to artificial intelligence everywhere around the world, private sector, public sector, international organizations, governments and so on. We have decided to define the ethically
aligned in AI in a broader way as trustworthy AI, but with a way, in a way they would have concrete reference to the legal system. So initially, the first thing that we said, is trustworthy AI has to be legally compliant, even before we start talking about ethics. Because laws are there, and we know how fluid and difficult to grasp is the digital subject matter, it is far from established that all the digital technologies out there easily comply with all the legal rules that we have in place. So legally compliant from GDPR onwards obviously, ethically aligned and then we define four key principles of ethical and responsible development of AI, which go from the protection of human autonomy and agency, to the prevention of harm, fairness, explicability. But we did also, and then we would, so we added a third pillar which is the robustness of AI. So trustworthy AI is also AI that has gone through some process, in an ongoing governance, not just an exemptive process, an ongoing governance that guarantees that best efforts are made to make this AI-product resilient and robust towards external attacks, right. Within the ethical
pillar we have done something that, in my opinion is a little bit of a game changer compared to the proliferation of ethical principles that are out there. We have tried to convert those principles into requirements, and the requirements into concrete questions. So to guide AI developers and designer and developers and deployers, through a process in a way that would enable them to self-assess at least, perhaps in the future to be assessed as regards their alignment with the good practices in trustworthy AI. Now this is the basis for the white paper on
AI that the European commission has presented in February, for what concerns the ecosystem of trust, and will be the basis for the forthcoming regulation on AI that will be presented in the first quarter next year [2021], although there are a number of problems there that we can elaborate on but obviously only if we have time. But still, the underlying DNA and overall texture of that regulatory intervention is in that initial input. The second is a set of recommendations for policy which are oriented towards data, towards the role of the public sector, towards skills, towards infrastructure. And them, they are very broad. I think they have been a little bit more diluted in the process of trying to get agreement between such a diverse group. Overall I think I have mixed feelings with respect to the experience of the High Level Expert Group.
I would save as a flag that we have been able to put the Ethics Guidelines on Trustworthy AI as being something that, at least has made a little step forward in the direction of something that is a huge castle, the regulatory framework for AI that we still largely need to build. Koen Smeets: I fully agree with you. And I was wondering, going back to Professor Coyle, how, what is your perspective on the regulation of disruptive technologies? And it's especially in the context of the earlier comments you made on the digital economy, and also the comments by Ms Stix and Professor Renda, and could you tell us more about especially the difficulties, for competition policy in a digital economy? What are its unique difficulties? Diane Coyle: Well, there are several things I’d want to say about that. First of all,
traditionally competition policy hasn't had to think as much about the dynamics of markets as it does in these digital markets. And that's because as Andrea was saying, the context now is that you're trying to make sure there's competition for the market so new entrants can get in, even if they then become dominant for a while, in the way that Facebook overtook Myspace, or other browsers overtook Internet Explorer. So there's a need to kind of reshape competition policy to think in that way, complicated by the fact that a lot of the big companies that we might worry about operate in many different markets. In a normal market inquiry or merger inquiry, you'd define a particular market that you're looking at. It's much harder when you're thinking about one of these very large companies that's got lots of different activities in a very complex ecosystem as it's called, of people supplying to the platform, people using the platform, moving into different markets with its userbase. The hard thing though I think is thinking forward.
There's a lot of debate now about whether Facebook should ever have been allowed to buy Instagram, at the time nobody was at all concerned about it because Instagram was very small. Was there anything competition authorities could have looked at at the time to give them a clue? Well, one possibility is you would look at the price that Facebook was willing to pay for Instagram. I can't remember exactly what it was, but it's a very large sum of money for a tiny company, and that was a clue. And then the other is that you need to look at, much more closely at the board documents and the strategy documents that competition authorities have access to. So it changes the way that you think about competition policy and apply it. But then the other thing is this whole question about regulation that Charlotte and Andrea have been talking about and I’d like to build on that. We've had this debate for a long time, framed by
business actually, that says regulation is bad, it clogs up the economy, it stops companies being as productive as they might be. Sure, there has to be some, but it's generally a bad thing and we want to avoid it. And there was a regulatory freefall really for digital companies right back in the 1990s. But you can also think about regulations and standards, and you know an example might be setting the voltage for electricity, which was about safety but also about setting a standard which created a level playing field, gave all the businesses in the market a clear set of standards that they all work to and grow the market. Another example
would be GSM and mobile technology, a standard set by Europe which you could also see as a burden on companies that were not using that technology originally. So regulations sets standards, shapes level playing field markets and makes it possible for them to grow. And there's also now a lot of discussion about what kinds of regulation do we need in these digital domains, which will range from mandatory codes of conduct, to particular technological standards and interoperability of data and so on. And I would point out that we wouldn't be having this debate about standards and ethics if companies had been behaving differently, and so part of what's happening now is a response to the behaviours. I don't think it's all about ethics, though. Ethics are really important but
being an economist I would say incentives are important, too. And a lot of you know, it's not that a lot of engineers and data scientists are bad people, they're not evil people, but they're operating in a system with very powerful incentives that shapes what they do. So public service procurement is an area I think is going to be interesting to think about. When governments are buying technologies in public-private partnerships or just to deliver public services, they need to think really carefully about what they put in those contracts. So at a minimum, data access. The data does not belong to the company that's providing the technology through a contract, it belongs to the public. I think it would also be really interesting to think about public service AI, and the public sector itself developing applications which are, because they're different business models then the kinds that operate in the private sector, will lead to very different kinds of behaviour, and I think it'd be very healthy to have those those comparisons. So an example might be smart cities, or transportation, or health,
where if public authorities can use their data, respecting people's privacy and data security, obviously, but use that to deliver benefit for people in general, then that is a form of competition with the private sector that will change the behaviour of the private sector. The other the final point I’ll make about this, is that we tend to have quite generalized debates about AI, about data, actually you need to get much more specific, because it's different in different sectors. A lot of attention focuses on the advertising-driven models, the big tech giants, and our personal data. There's a lot of other types of data as well and we need to think about how the value from those gets distributed. One of my favourite examples is John Deere, the tractor manufacturer, which has provided lots of sophisticated IT-equipment and software in the cabins of tractors, and farmers get a lot of useful information from that about the soil, and the weather conditions and so on. But John Deere encloses that data for itself and is creating new
software and new services to sell, which are higher profit margin than selling tractors. And so it's capturing that kind of value, rather than sharing that with the farmers and actually in that case is even trying to use US courts to forbid farmers to mend their own tractors, on the grounds that they have copyright over the software in the cabins. And so that's an another kind of area that's not about personal data and we need to think much more about the Internet of Things and, particularly in Europe where there's a industrial advantage in that kind of technology.
Koen Smeets: So I think that's a very, very interesting, comprehensive answer on the difficulties related to this, and I was also very curious, because you've also been quite critical on how this relates the way we teach economics. Because economics, especially in its basic models, it's quite negative on regulation, it's assumptions say, markets are perfectly competitive and work and such. So I was wondering, and you also have talked about these assumptions influence policy makers and economic experts in the society, so could you expand on those subjects? Diane Coyle: I’ve been very involved in trying to change the economics curriculum over the past few years. And it's driven by an experience with policymakers who learned their economics some decades ago and it was very much shaped as, by this perfect market benchmark, and so always starting from the position that generally the markets are the best way to organise things, but you might think about exceptions, you might think about market failures. And although I think that's a really useful intellectual framework for testing how you might make a more efficient use of resources, I don't think it's the right place to start in the modern economy where things like increasing returns to scale, market power power, structures in the labour market are, there's such obviously an important empirical phenomena that you ought to be starting there. And of course you can then use the underlying economic theory to, as a kind of thought experiment or to test your ideas, but you should be starting in a different place. And so I hope that, I think that's changing. There's huge student interest
in anything about digital economics, for the obvious reason that it's really affecting our lives in in a big way. And so as I do detect change. I wish there was more economic research going on in this area, I mean obviously some fantastic academics, but it's been a bit slow. I started writing about digital technologies in the 1990s, we've had the internet widely available since at least the mid-1990s, and it's really only just now that you're seeing a huge growth in the amount of economic research being done on competition in digital markets, and I regret that it's been so slow. We should be we should be further ahead
than we are in terms of both the analysis and the data, the empirical understanding. Koen Smeets: I saw professor Renda quite nod, could you also expand on how you see this? Andrea Renda: Well I was nodding on, first of all on Diane's example of John Deere because I think it's very telling of what is happening in a number of markets, and what has already happened in a number of markets in particular and economic sectors, in particular in the United States, where farmers indeed need to purchase access to data coming from their own land. This is something that transforms them into slaves to those players that are able to capture the value from the real economy activity that they perform. Indeed, this is exactly the concern that has
led, the way I interpret it at least, the European Commission to launch this data strategy based on two main pillars. One is the, it's a sort of foresight, it's a vision of the upcoming evolution of digital technologies, in particular from the centrality of the cloud as the place where we store data, to more distributed, even ultimately decentralized architectures, where not only we store the data more locally, we avoid sharing the data widely, and we apply artificial intelligence in a more, at a more local level. One example, autonomous vehicles cannot afford shipping the information to the cloud whenever a decision has to be made and then receiving it back after the cloud, the cloud-based artificial intelligence has elaborated that information. This creates latency, creates connectivity costs, it creates some security problems, because everything that travels long distances is potentially exposed to attacks. In principle, ideally, we would have a big brain inside the autonomous car, but at current technology, a big brain autonomous and able to fully process all the information in an autonomous car, means a half-hour battery duration, right. So, technology advances, currently we are at the at the situation in which a lot of artificial intelligence and data storage can be put in what we call "the edge". It's an intermediate layer between the things, the
connected things, and the cloud, and edge cloud architectures are much more, let's say prone to data management by real economy players, car manufacturers, farmers, energy companies, and if you create a governance and a legal framework that it is conducive to such a sharing of data between the players and the producers that populate those sectors, you might create the preconditions for their stronger bargaining power vis-a-vis the tech giants, and a bit of a rebalancing of this value that has been captured by just a fistful of players so far. So that is a very acrobatic attempt, but the Data Governance Act, the Data Act next year [2021], the potential scaling up of GAIA-X as a Pan-European project with a federated cloud environment, is resting on this idea. And a second idea which I think is very interesting for economists and people that study social sciences and decision sciences more generally. The idea that, maybe it's finally Larry Lessig's time. Meaning, Larry Lessig in the mid 90s and late 90s, wrote about the the prevalence of code as, rather than law as being the determinant of what is possible in the internet environment. Well the attempts through GAIA-X and data spaces of the
European commission can be interpreted as a way to translate legal codes into software codes. Meaning, being part of GAIA-X means, in principle, we'll see the realisation in practice, committing to compliance with GDPR by design, and committing with some forms of data interoperability by design, and perhaps implementing protocols for use and control over data by design. So we're actually thinking, you know the normative power of Europe that we've been thinking and then Charlotte was mentioning before, now the idea that there's a Brussels effect, or that Europe can be a standard setter also for digital technologies around the world, in my opinion it's chiefly dependent, at this moment, on Europe's ability to create an environment in which it governs technology also by technology, not only by legal rules, and courts, and regulatory agencies, and that I think is an enormous area for economists, for interdisciplinary social scientists to study at the moment, and it really increases and strengthens, if you wish, the muscles of the of the economist if the economist is willing to venture into this, into alternative forms of governance, which come from a long tradition in economics, obviously, and their interaction with technology. Koen Smeets: It's very, very fascinating. Ms Stix, could you expand on that as also from
the technical perspective, and perhaps if you also have comments on the economic perspective. Charlotte Stix: I mean sure, I mean, so the sort of products that we produce and deploy in Europe and our competitiveness, or the competitiveness of our industry, is intrinsically linked to ethical and technical as you said, considerations and I mean this has been mentioned before by Diane I think, ethics is not you know the be-all end-all, but ideally ethics and technical considerations should merge and align, right. And in Europe they do that, and that is a really important direction to point out and in the white paper which follows the ethics guidelines, the seven key requirements from the High Level Expert Group that Andreas mentioned.
In the white paper that has been translated into technical obligations, or legislative obligations for high-risk AI-systems which is equally grounded in technical requirements, you have requirements for training data, so to ensure that reasonable measures have been taken, that outcomes don't lead to prohibited discriminations, data record keeping, documentation programming and training, information provided, robustness and accuracy, so to ensure how a system can adequately deal with errors in the life cycle or through attacks and human oversight. And I think those sort of mixtures really do put Europe in a unique position in comparison to other governments, internationally speaking. And it also could empower European industry. So yes, there is a Brussels effect, but it could actually also lead to novel innovation quite frankly.
So the testing and experimentation facilities that will eventually need to be built, in order to ensure adherence to these ethical/technical obligations, that ensure that you adhere to the relevant legal framework will set completely new structures as to what products that come onto the European market will look like, and that can encourage innovation because a lot of these considerations are actually forward-looking and they address both technical and societal problems. So if you think about the long-term effect of AI-systems on the climate, yes it is often touted as, you know being able to tackle climate change but it is also a massive contributor to worsening climate change. So if you think about, for example putting this as one of the technical obligations to address topics such as these, you could encourage the European industry to focus on energy-efficient learning, which might put Europe into a different position, a different position in the global scale. And as Andrea has mentioned you know, Europe does have the data strategy, does think about edge computing, GAIA-X is an initiative across different member states, and there is value to capture there. And as Diane said you know, with the example with the farmers,
that is a really big problem that Europe is also addressing and trying to pre-emptively tackle. So it is suggesting to open these forms of environmental data, in order to harness them for the individuals and for the public sector and so that they are not, you know resold for purposes or for groups of people that should have access to these things in first place. So I think Europe is really going into a lot of different directions here, trying to mix and merge competitiveness with ethics, with technical considerations, under the helmet of pushing the ecosystem that we have from an industry perspective and the ecosystem we're creating from legislative perspective. Koen Smeets: Fascinating. Professor Coyle, would you care to expand on professor Renda's and Ms Stix' comments? I think you're still on mute. Diane Coyle: Sorry. It's the phrase of 2020, you're still on mute. The thing that struck me actually listening to Andrea and Charlotte is, I
completely agree that there is a real opportunity for Europe here, not just to shape outcomes in Europe but to shape them globally, and to have a leadership role in setting standards, and providing models. The thing that struck me though is the need for an interdisciplinary approach. And this applies to academics working on these things as much as it does to policy makers. You need computer scientists obviously because you need great technical know-how to set standards, regulate effectively, deliver value for people. Economists, lawyers, deeply involved in competition policy and writing regulations. You also need to involve politics, because these changes in our society that are coming about, they need to have legitimacy. And we know we're in a context of great polarization in lots of countries, inequalities being exposed and broadened by the pandemic and the economic crisis. And it's going to be really important, because the
technology will drive significant social change, to have political legitimacy and accountability, and then also social psychologists, behavioural psychologists, because this is all about how people, how people behave. But if you think back to the Industrial Revolution, we tend to talk about something like railways as changing transportation, which obviously it did, but it also drove urbanization because food could be grown outside cities and brought into cities and so the population that could be sustained in modern urban centre was much bigger. And that's been a huge change in social, political, and economic life through the 20th century. And that's the kind of scale of eventual impact on our societies that we're talking about with any general purpose technology like digital and AI.
And so fundamentally, although I think what we've been talking about is important, the really important thing is the legitimacy of the changes, and ensuring that these technologies deliver benefits for everybody and not just making a few people very rich indeed. Koen Smeets: And focusing on the economics curriculum, you mentioned the importance of interdisciplinarity and I think we see this also not in this interview but also in the other interviews in this series, and I was wondering how, what should we exactly then change in the economic curriculum? And also, should we include more pluralism, should we include more real-world perspectives, should we include more interdisciplinary, how do you see this? Diane Coyle: Well, the example I point to the CORE economics textbook, the economy, which I was one of the co-authors of, and I think it does take this much more empirically-founded approach and incorporates things like power dynamics, and inequality, and distribution right at the heart of the curriculum. I think it is important to understand older debates in economics, but I wouldn't go so far as to say that, like some of the humanity's economics is always about contesting sets of views and values, it's both. Obviously, we all bring our values to these questions and it has been a mistake in economics to say that you can separate the normative and the positive, the values, choices, and the empirical evidence. I don't think you can separate them, but at the same time I think it's really important that as economists we try to be as impartial as possible, looking at evidence and bringing empirical evidence to bear on these social problems. So it's
an uncomfortable middle position, but I’m not an advocate of complete pluralism in the curriculum. Koen Smeets: That's interesting. Ms Stix, how do you think we should include in the economic curricula, what would be most important from a policy perspective and perhaps also an economic perspective? Charlotte Stix: I mean, I think I can just echo what has been said so far. I think in all fields,
not just exclusively economics, if you start talking about emerging technologies it is really important to include a lot of different fields working on this. And coming back to the earlier point, I think particularly technical researchers, it's important to understand what you are looking at and the actual capabilities of this technology, not as it is now, when it's already on the market, but what it can do in two or three years. You really do need to speak to those researchers doing their PhDs right now on these topics, in order to know what is the cutting edge. Well, because these technologies shape economic markets so much and they shape government's decisions so much and so quickly, that you almost need to have an anticipatory view. And you can only have this anticipatory view if you don't react but if you sort of already understand what the next steps and what the timelines for technological development will be. Now that doesn't mean you need to become a technical expert, by no means is that. Well, it might be possible but I don't think it is required,
but it does mean that you do need to engage with those people that are working on these technologies right now. Because otherwise the cycles are becoming too quick and you can lose track really quickly and it does shape your research or your proposals in your economic work. Koen Smeets: And then as a closing question before we move to final statements, Professor Renda, how do you see this in light of professor Coyle's and Ms Stix' comments? Andrea Renda: Well I'm, I agree with with both of them. I’ll bring in a little bit
of personal experience as well, in trying to encourage students of economics today to really listen to what their teachers have to say, but at the same time develop their own intellectual path, independently and in a multidisciplinary way, as much as possible. I studied economics, I specialised initially in a subject called "economic analysis of law", or "law and economics", which was heavily dominated by the Chicago School and Neoclassical, Neoclassical economics behind heavy use of very standard cost-benefit analysis, the translation of this into an approach to competition law which I would say was almost minimalistic. And I have navigated through those waters by trying to stick to my own understanding and my own beliefs, which were very sceptical, and I was very sceptical of many of those principles. And I kept applying this, and I still apply this, when today for example, we apply economics in public policy in a way that still uses maybe very standard tools like cost-benefit analysis that are in most cases unrelated or disrespectful, in some cases of governance, of distributional impacts, and stick to something that in my opinion is one of the, of the key problems but has been one of the key distinctive traits of economics over the years, which we call methodological individualism.
In economics largely, was still, analyse people's utility or happiness, in a way that is completely unrelated to what happens in the surrounding or what others have, so the relative dimension of that. And this I think has made economics say a science, that princes and policymakers looked at, as a rocket science, which it's obviously not, and so it has determined part of the popularity of economics over time. But at the same time, it's a huge limit in this, and this social science and something that calls for contamination in the positive sense, with many, many other social sciences today. So, in principle, I would ask students today and I would ask the ones that develop curricula in economics to try and depart at least partially from what we normally have, especially in textbooks microeconomics books, such as indifference curves or things like the utility functions, and what we immediately start learning, very, very soon in the economics 101, which is the fact that people's happiness and utility can easily be proxied by income. And that is something that in my opinion has created disasters in the
application of economic policies, in developed and developing countries around the world. Koen Smeets: I’d had loved to expand on that but we're already quite a bit on time. So for closing statements, Professor Coyle, perhaps also in light of what Professor Renda just said, if there's a one thing you'd could say to students in economics watching today, related to the topics we also discussed today, what would that be? Diane Coyle: My one piece of advice would be about the kinds of questions that you pursue in your career, in your studies, and your subsequent career, wherever that is. There are very strong incentives in life to stick to small problems, you know fix a particular detailed policy issue or research something that will get you a paper published in one of the economics journals. There are lots of really clever people spending all their energy on small questions. We've got some really big questions facing us at the moment, and so my advice would be, obviously students are really interested in those big issues, and so have the courage to pursue those big issues, because it needs the younger generation to be working on them. Koen Smeets: I think that's a beautiful answer. The same question, Ms Stix,
if you, if there's one thing you could say to students, especially in economics watching, related topics we discussed, what would that be? Charlotte Stix: Sure, I mean, I think I would pretty much, well first of all agree with what has been said, and come back to what I’ve said before about working with various different experts. You can't work in silos anymore in the world that we're in, not with this technology either. And you cannot come up with completely novel ideas focusing only on your, you know narrow, specific question that you're looking at, it's really important to broaden your horizon and to engage with all of the knowledge that is out there. And you know, the knowledge from frankly
various different fields, and I think that's where fruitful connections, and new ideas, and also approaches to tackle specific problems can be drawn from, and that's really important. Koen Smeets: Yes, I fully agree and I hope that this interview-series also can contribute to that. Professor Renda, for you the same question as the very last closing questions. Do you have any last tips, recommendations, advice? Andrea Renda: Well I think I can maybe relate my answer back to what has been said throughout the hour, I think to digital technology in particular. I think the emergence of digital technology
and interconnected environments, such as the internet, has given economists and social scientists more generally, a unique opportunity to study the evolution of an ecosystem, that could be seen, at least at least initially, could be seen as a standalone one. Still today I think, we have the possibility of studying the evolution of the digital economy as a as a living system, if you wish. Where the needs, the external needs, the needs to perform certain functions and technological evolution, determining a different morphology of that living system over time. We've seen some basic foundational elements which Diane has summarised in her first answer, we've seen the first evolution, which largely was due to the way in which the internet was structured you know. Code determines what's possible. With today we are building from the central nervous system of the internet, the cloud, we are building the peripheral nervous system. The potential that we will have in using the Internet of Things for regulation, for public policy, for social life, for the economy is enormous, and we need creative minds that are grounded in social science, including economics, that help us understand and anticipate, what this will mean in terms of governance, regulation, and public policy. So be applied,
be creative, be broad-minded, and importantly, be inspired by the public good, which I think is very important and sometimes difficult. For, and especially economists that have a very rich market in front of them, if they specialise in something else than the public good, to remain really concentrated on what still economists can do that is great today in terms of contribution to the evolution of our public policies, and overall the way in which we govern our economy and society. Koen Smeets: Thank you, I think it was a beautiful answer and a great closing statement. And I want to thank each of the panellists for taking the time today,
and we hope to see the viewers again at the next episode. Thank you!
2021-04-27