Skip to content
Department of Law

EUI professor advises the EU on improving regulation through use of data and AI

EUI Law Professor Giovanni Sartor recently published a detailed report on how the EU's regulatory process can be improved through better use of data and artificial intelligence in impact assessments.

20 September 2022 | Publication - Research


The in-depth analysis by Professor Sartor, entitled The way forward for better regulation in the EU – better focus, synergies, data and technology, was commissioned by the European Parliament’s Policy Department for Citizens’ Rights and Constitutional Affairs, at the request of the JURI Committee.

The report delves into topics such as the role of data in the policy cycle; costs and benefits of data collection and use; re-use of collected data; choosing appropriate evaluation methodologies; and complexity and uncertainty in assessing policy impacts.

Fundamentally, Sartor finds that information technologies, and in particular artificial intelligence (AI), can enable a more extensive and beneficial use of data, and that the use of data in ex-post evaluation can improve the regulatory process.

In a brief round of Q&A, we put these questions to the author.

Q: In which policy areas do you see the most exciting prospects for AI and data technology?

A: There is a vast scope for using AI and data technology to know more about the effects legislative instruments have had, or will have. These technologies would also enhance quantitative assessments of policies. It seems to me that AI has been under-used in planning legislation. A lot more can be done, both using analytics (extracting information from data) and through simulation (building 'digital twins' of social realities, such as marketplaces, to see how they respond to changes). Regulations affecting physical environments and those concerning the digital economy are the two areas where technological support can really increase the effectiveness of impact assessment.

Q: Your report stresses the need for evidence-based regulation. Do you have examples of regulation that turned out to be insufficiently evidence-based?

A: According to OECD, the EU is doing well compared to national governments. However, there is still room for improvement. For an idea of the quality of the regulatory assessment, we may look to the evaluations provided by the Commission’s (independent) Regulatory Scrutiny Board. It has found up to 40 percent of the evaluations by the Commission to be inadequate, and in some cases, the resubmission has also been considered inadequate. We shouldn’t judge too harshly the efforts so far, since assessing impacts is a very difficult task and considerable progress has been made.

Q: Do the European Parliament and the Commission and national governments see eye-to-eye on this issue?

A: It seems to me that the Parliament and the Commission agree on the importance of the assessing expected impacts of legislation and evaluating what has been delivered. The Parliament rightly focuses of ex-post evaluations, which should be done much more extensively. Ex-post evaluation would enable Parliament to exercise critical control over legislation and take initiatives to promote revisions. There is also the need for stronger collaboration, so that the Parliament can intervene before the enactment, in a timely way, taking into consideration the assessment by the Commission. Finally, the Parliament also needs to have adequate internal skills to assess the evidence provided.

Q: One of the first recommendations of the report is for the public sector to catch up with the private sector in terms of data collection and 'crunching' capacity. How realistic a goal is this for the public sector? Aren’t there normative and resource-related constraints on rapid implementation of AI and big data?

A: The constraints are huge. Aside from legal constraints, in particular data protection law, there are constraints related to the fact that a lot of data are controlled by private entities. But there are also opportunities – in particular, relating to the vast amount of data that is controlled by public bodies, and which can be gathered through the exercise of public functions. I think that the normative constraints can be overcome, for instance by extensive anonymisation-obfuscation of the data used for impact assessments. And private companies should be required to provide anonymised data to be used for public purposes.

Q: How does your report enhance the Digital Markets Act and the proposed Digital Services Act?

A: The Digital Markets Act and the Digital Services Act provide the scenario in which we can rethink the issue of the access to data for public purposes. This is relevant in two ways: data are needed to improve public policies in general, and – more specifically – public institutions need data to effectively regulate digital interactions.

Last update: 20 September 2022

Go back to top of the page