Republished with full copyright permissions from The Washington Daily Chronicle.

The recently published report by the U.K.’s Competition & Markets Authority (CMA) has sparked important discussions about the potential monopolization of the artificial intelligence (AI) industry and the associated risks to consumers. While it remains uncertain whether these scenarios will materialize, policymakers are grappling with the complex task of effectively governing AI, which holds immense potential for commercial advancement as well as potential risks. The report’s findings offer valuable insights into the regulatory concerns surrounding AI, particularly in relation to foundation models, which form the basis of various AI technologies.

Foundation Models and Barriers to Entry:
The CMA report places a spotlight on foundation models, a type of AI technology trained on extensive sets of data and adaptable to a wide range of tasks. Concerns are raised regarding the increasing computational power and data requirements for training these models, which may inadvertently create barriers to entry and reduce competition in the industry. Professor Anton Korinek of the University of Virginia emphasizes the importance of recognizing the need for a competitive market that fosters innovation to avoid concentration at the frontier of capabilities.

Recursive Self-Improvement and Speculative Scenarios:
The report also addresses the concept of “recursive self-improvement,” whereby those at the forefront of developing foundation models may utilize their capabilities to further enhance their models, potentially distancing themselves from competitors. While speculative, this scenario poses potential risks to fair competition and further complicates the regulatory landscape. Statements from industry leaders have already highlighted the possibility of a winner-takes-all dynamic, underscoring the urgency for proactive measures.

Open-Source Models and Balancing Public Safety and Competition:
The report considers the impact of open-source models on market power abuse and consumer protection. Maintaining powerful models as open-source may reduce the ability for AI developers to exploit their market dominance, offering potential safeguards for consumers. However, the tension arises when considering the possibility of harm that could be caused if powerful AI models were more widely accessible. Policymakers need to strike a balance between promoting competition and ensuring public safety, as powerful open-source models are currently developed outside the U.K., limiting the regulatory influence.

Accountability and Consumer Protection:
The report highlights the vital importance of holding companies accountable to protect consumers effectively. The Federal Trade Commission’s investigation into a prominent AI developer further underscores the need for proactive consumer protection measures. Collaboration among competition regulators and the development of coordinated efforts to prevent firms from evading regulations is crucial. The proposed European Union’s AI Act, with its transparency requirements and risk assessments, imposes certain obligations on companies behind foundation models, aiming to instill responsible practices.

Cautious Approach and Future Actions:
The CMA’s report acknowledges the need for a cautious approach and emphasizes a “significant program of engagement” with stakeholders. The regulator’s stated intention to update the report in early 2024 affirms ongoing commitment to evaluating the rapidly evolving AI market and its impact on competition and consumer protection. Additionally, the legislative bill in progress, granting the CMA new powers to regulate digital markets, could potentially be extended to AI developers found to abuse market power or harm consumers.

As the AI industry continues to evolve dynamically, policymakers and competition regulators are faced with the complex task of fostering competition while ensuring consumer protection. The CMA’s report provides valuable insights into the challenges and potential risks associated with foundation models. It calls for proactive regulation, effective coordination among competition regulators, and a careful balance between promoting competition and safeguarding public safety. By collaboratively addressing these concerns, the AI industry can thrive while ensuring responsible development and deployment of foundation models for the benefit of all.