In the fourth episode of our podcast series Competition Law in Digital Markets in Latin America “Competition Law and Innovation II”, Thibault Schrepel covers up the most groundbreaking development of the last decade, Artificial Intelligence.
We start by discussing some issues involving the Digital Markets Act like the gate-keeper concept, the use of data in AI, self- preferencing and the network effects in order to answer how the DMA relates to competition and innovation in digital platforms.
Secondly, Dr. Schrepel explains the dynamics of Artificial Intelligence from the inside but most importantly its dynamics with the outside world as an example of how today´s markets are evolving and the necessity of competition agencies to reshape their traditional approaches.
Last but not least, the use of AI by these agencies is put on the table as a way to better understand their own case law, evaluate the impact of how they are doing merger control and even detect anticompetitive practices in the markets.
“I think is not a good idea to take the DMA and try to transplant it such as it is in other countries. In Asia wouldn’t work, in Latin America wouldn’t work, and in the United States wouldn’t work.”
“The reason for that is because if you look at EU markets you see that there are specificities that you may not encounter in other regions of the world.”
“Open AI or other related companies won´t be covered by the DMA which it already tells you something about the fact that the DMA is very rigid and cannot easily be adapted.”
“When we talk about AI, we talk about dynamic competition and something that would reshape markets. Now once you settled that, the role for the policymakers and specially enforcement agencies, is to make sure that it can develop as it should.”
“The problem is that this kind of ex-ante regulation cannot adapt. So, if they prove to be ineffective, is hard to go back to the Congress, and just say “let´s vote to another version of that, even to revise”, it usually takes a lot of effort.”
“If on top of being ineffective, they prove to be effective and actually creates negative effects on the market, then you are facing the same issue that they cannot adapt.”
“In AI you already see that these kinds of provisions that you may always want to enforce in the same way, may be good sometimes and some other times may actually freeze the competition and I´m not sure this what you want within your AI ecosystem.”
“I think what is key for AI companies’ growth is access to users. Because you can learn from what users are doing with your system which is highly valuable.”
“To sum up, regulation may be one day but first, it may be ineffective, second of all, is hard to adapt and when you are trying to regulate the systems, that’s on the daily basis if not on the hourly basis so enforcement may be best.”
“When it comes to enforcement it doesn’t mean you can enforce everything at once and I think the priority should be practices freezing the ecosystem because this is what you want to avoid.”
“Competition agencies can also use AI, those enforcers can actually use AI to better understand their own case law, what they been doing, take a step back and maybe redirect/change their policies. They can also use AI to better detect practices”.
“Why not use machine learning based on your past experience enforcing merger control and then being able to absorb what happened after your decision on the markets to try to better shape the way you do merger control.”
“As a policymaker I would say: be interested in those issues but also
The Digital Markets Act (“DMA”) set a precedent for ex-ante regulation in digital markets. It is essentially a European centric regulation. It is based on the case-law that has been developed in Europe enforcement.
Many of the conducts described in the DMA are considered as always harmful to competition in the markets. But the DMA required first to classify a company as a Gatekeeper. This concept has been subject to debate as the existence of a plurality of Gatekeepers means that Gatekeeper is not possible. Likewise, if these are practices that are always harmful should not them be generally banned by competition law?
As we have discussed in prior episodes, digital markets dynamics are constantly changing. Competition Law generally has been a framework which has been now because it can adapt to different industries and market conditions. On the key questions regarding the DMA is if it can be adapted to these changing circumstances.
Law making is a complex process which requires a long process. Creating such legal frameworks may not be suitable to address the challenges that digital markets require. The positive or negative effects of such regulations are yet to be measured not only in digital markets, but markets and companies which use such services in their business and product innovation.
Lawmakers, regulators, and enforcers can look into technology and AI as tools to tackle the challenges in digital markets. These tools can help them better understand the dynamics, propose compliance actions, and use the existing legal framework and competition law toolkit to achieve the purpose of competition law, innovation, and consumer welfare.
Associate Professor of Law at the Vrije Universiteit Amsterdam and Faculty Affiliate at Stanford University
Subscribe to our newsletter! Get notified about updates and be the first to get early access to new episodes.
