I. My risk regulation story

This year marks the twentieth anniversary of my first encounter with risk regulation. This long ago, I was one of the lucky students enrolled in the (then brand new) MSc in Regulation at LSE. A team of young professors, which included Julia Black and Rob Baldwin, had put together a “core course” in Law and Politics of Regulation. In this course, we dealt with such essential and cross-cutting questions as “why regulate?” and “how to regulate” in different settings. Other courses gave us the opportunity to discover environmental regulation, financial regulation and many other sectoral or specific perspectives.

Prior to joining this programme, I had read law and economics (separately) in France (integrated law and economics programmes were not offered at the time) and surely the first cause for my enthusiasm for the study of regulation was the genuinely interdisciplinary approach embraced at LSE. The dialogue between disciplines which studied different aspects of a same phenomenon was precisely what I had missed so much at the Sorbonne.

Along my academic path, the study of regulation was where I was first able to connect economic and legal dots while at the same time realising there were many more dots to the picture. I came to appreciate that the views from political science shed light on issues of institutional design. I was fascinated to learn that tools from linguistics could be mobilised to analyse the regulatory conversation (between regulatees, regulators and courts) and draw attention to aspects which would not otherwise have showed up on either economists’ or lawyers’ radars. Importantly, I grasped that many other sciences, in particular those which help identify and assess risks, were part of the regulation stories.

My core interest has since always been the dialogue between law (which ended up being the flag discipline of my small academic boat) and other disciplines. One focus in my later studies was how legal institutions – be it courts, governments, EU institutions, competition authorities, regulators – dealt with expert evidence. Methodological issues regarding how judges and administrations handle numbers, hypotheses, probabilities, uncertainty, conditional conclusions are relevant to all sectors. This is a topic where specialised institutions, but also experts and scholars can learn from other fields and where a journal such as EJRR can help cross-fertilise regulation studies. For this reason alone, in addition to looking forward to working with distinguished colleagues, some of whom have become friends, I am honoured and happy to join the board of editors and am immensely looking forward to opportunities to discover new analogies and refine my understanding by learning from experience in regulated sectors I do not know well.

II. Defining risk regulation research

Over the years, my interdisciplinary curiosity has largely taken me to subjects other than risk regulation and I am not current on the risk regulation literature as it stands. I hope to bring to the journal the perspective of a law and sciences scholar who still knows just enough about the study of regulation to establish links with law and economics or behavioural analysis of law developed in areas other than risk regulation.

III. What is the future of risk regulation?

Risk regulation studies are bound to expand in scope as new risks emerge. Applying existing analytical tools to new problems will also, in all likelihood, lead to a more refined understanding of regulatory processes. In other words, the field of regulation will grow in both breadth and depth.

Increased breadth will come from new risks. Among them, the risks stemming from the generalised use of Artificial Intelligence (AI) strike me as particularly worthy of scholars’ attention. Those risks have not yet been fully identified, let alone assessed, and managing them is probably a long way off. Yet, one cannot help but think that AI will bring a new cluster of risks. Self-driving cars are probably one of the easiest examples, next to domestic robots, which are the next bots in line in our changing everyday lives. In themselves, the risks these devices pose are not new: we have known road accidents ever since cars have been around. We have known domestic accidents since (at least) stoves and other home appliances have been around. There are systems in place to prevent both road and domestic accidents and manage their consequences when they do occur.

The new questions, which currently prevent wider dissemination of driverless cars and domestic robots (besides some remaining product development issues) “only” concern who should be responsible: the seller of the car, the manufacturer, the programmer, the company offering the transport service, the passenger? A core difficulty, as in other areas where AI is applied, is that the entity which controls the car is not a person (legal or natural) but an algorithm, more precisely an evolving algorithm that teaches itself how to learn from “experience” (data). It could be that, at least in some cases, what causes an accident is a faulty learning process, but it would not be helpful to hold an algorithm responsible.

Beyond the issue of liability, how can risk prevention be organised? Can one regulate algorithms so that they become better at preventing risks? If such regulation is possible, how would it be enforced? Risk regulation scholars will have to consider whether algorithms could regulate algorithms. Issues of institutional design would take a whole new turn (meta algorithmic design?). In terms of interdisciplinary mix, a computer scientist will surely have to be brought on board, along with economists, lawyers, political scientists, sociologists, anthropologists and linguists. As machine learning finds more and more applications, these and other questions pertaining to the regulation of algorithmic risks will only gain prominence and I trust and hope that regulation studies will branch out towards IA.

In addition to broadening their scope, regulation studies (including risk regulation studies) will also gain in analytical depth. I expect this will happen as regulation scholarship further integrates the teachings of behavioural sciences. At present, models of regulation still come from neoclassical economics and the rationales for regulatory intervention are still typically presented in terms of market failure. Through this notion of market failure, regulation scholarship partly embeds rationality assumptions which characterise neoclassical economics and which have been shown to be inaccurate.Footnote ¹ While of course not all regulation scholars assume that regulators, regulatees and citizens are all perfectly rational, incorporating insights from psychology more systematically is a scholarly endeavour which has only just started. It will continue to have a bearing on our understanding of when and how best to regulate. In the field of risk regulation, scholars have long been aware of the gap between risk perception and expert risk evaluation, but it was only in the last decade that they started drawing policy implications from behavioural insights on attitudes towards risk. Both studies on judgment about risk and choices in risky environments are of the utmost interest to risk regulation scholars. Behavioural risk regulation scholarship has been fast growing and it still has a long way to go.Footnote ² Just as other fields of law and public policy, the analysis of risk regulation deserves to be revisited in light of insights from psychology,Footnote ³ and I trust that, in this regard, risk regulation studies will be leading the way for regulation studies in general.