Ensuring competition in AI will also preserve democracy, experts say

9 months ago 32

With artificial intelligence, focus on enforcing competition laws is vital and goes hand in hand with the preservation of democracy, industry experts said at a panel discussion in the European Parliament on Monday (19 February).

In 2023, the German competition authority (Bundeskartellamt) looked at the artificial intelligence industry, especially the Microsoft OpenAI partnership. In January, the European Commission decided to look into competition in virtual worlds and generative AI and opened calls for contribution.

The French competition authority (Autorité de la Concurrence) also took the initiative to look into competitive practices of the sector in February.

“We [should] rename competition authorities or add a democracy competition competency in their names because this is what it is all about. Competition is also democracy protection,” Tobias Haar, member of the general counsel of generative AI startup Aleph Alpha, said on Monday.

Barry Lynn, executive director at the non-profit organisation Open Markets Institute, warned that AI amplifies existing digital downsides.

In his view, the EU should change its competition lenses “to focus on harms to democracy and human liberty”. From this perspective, one could “see that behaviour regulation of these corporations is most important”.

Of paramount importance is to limit Big Tech “ability to engage in manipulative, distortive, discriminatory behaviour,” Lynn said.

Andreas Mundt, president of the German competition authority, similarly said that, when it comes to artificial intelligence, one “cannot overestimate the risks for competition, for society, and for democracy in general”.

His French counterpart, Benoît Cœuré, pointed out that the EU got it right with AI, as the Parliament and the Council “started thinking about the risks,” an approach taken within the EU AI Act.

“It’s more important than competition, in a way,” acknowledged Cœuré.

Still, the French competition authority president shared his concerns at the fact that it took years for competent authorities to apply the EU data protection law (GDPR), and he was therefore worried how much time national authorities will require to enforce the AI Act.

Talking about AI as a fast evolving technology, the EU Competition Commissioner Margrethe Vestager said: “Is the AI revolution going to follow the same history as web 2.0? I have very serious doubts”.

Contrary to 2000s, when entrepreneurs were “exploring uncharted territories,” the current technological change is “not happening in a vacuum”, she explained.

Indeed, AI models development depends on data, cloud capacities, chips, and specific skills that are already controlled by a handful of tech companies, which Vestager concluded creates “huge barriers to entry” for AI startups.

Carel Maske, director of competition at Microsoft, suggested that the AI industry was not that concentrated.

In his view, the AI value-chain contains five layers where different actors compete. These are land and financial resources, infrastructure (data centers and the cloud), AI models, AI systems, and distribution channels layers.

Maske mentioned “Microsoft partnering with OpenAI” – which the European Commission has said it would look into – saying that “there is no exclusivity in Microsoft using that technology”.

Yet, Germany’s Mundt argued that although Microsoft’s relationship with OpenAI was a “soft cooperation,” there is still an influence by Office’s parent-company on the world-renowned ChatGPT creator company.

Mundt announced that his organisation will support the European Commission in its evaluation of competition in AI.

“I can promise you, we will look at all these corporations very carefully, and if it only gets close to a merger, we’ll try to get in,” he stated.

Maske said that the Microsoft-OpenAI partnership is “not something under the radar” of competition laws, but Lynn of the Open Markets Institute was not reassured, pointing to the way Big Tech companies do business.

He suggested there is a need for “bold solutions” – to separate AI startups and Big Tech partnerships, impose a ban on price and information delivery manipulation, break cloud industry monopolies, and take away personal data ownership from Big Tech.

Mundt said the EU’s hope against Big Tech lies in small an medium-sized companies, which own very high quality specialised data, which “can be used to build specialised AI models [that] could outperform foundation models in their respective domains”.

“In a nutshell, Big [Tech] AI tries to do everything well [while] small and specialist AI could maybe do one thing extremely well. This is Europe’s big chance,” concluded Mundt.

[Edited by Zoran Radosavljevic]