The U.S. Department of Justice and Federal Trade Commission, in tandem with their fellow competition law enforcers from Europe (the European Commission, or EC) and the United Kingdom (the Competition & Markets Authority, or CMA), issued a joint statement on July 23, styled “Joint Statement on Competition in Generative AI Foundation Models and AI Products.”
This Joint Statement touts enforcer’s concerns with theoretical competitive harms from AI foundation models, which are “artificial intelligence models that are trained on a broad set of unlabeled data that can be used for different tasks, with minimal fine-tuning.”
The Statement, however, is highly problematic. It gives short shrift to the substantial economic benefits of AI and provides and misrepresents AI’s competitive risks and gives short shrift to the substantial economic benefits of AI.
Unless it is disavowed by American enforcers, the Statement will at the margin reduce private investment in the adoption and deployment of AI, slowing innovation and harming the economy. Furthermore, its negative effects will fall disproportionately on U.S. firms, which are leaders in AI. This will undermine U.S. international competitiveness.
The Joint Statement Misrepresents AI’s Competitive Risks
The Joint Statement focuses on a variety of potential competitive threats arising primarily from huge high tech firms (such as Google, Facebook, Amazon, and Microsoft). It specifically is concerned about:
MORE FOR YOU
Concentrated control of key inputs. These include “[s]pecialized chips, substantial compute, data at scale, and specialist technical expertise.” This could put “a small number of companies in a position to exploit existing or emerging bottlenecks across the AI stack and to have outsized influence over the future development of these tools. This could allow a small number of big companies to shape innovation “to their own advantage, at the expense of fair competition that benefits the public and our economies.”
Entrenching or extending market power in AI-related markets. Foundation models purportedly benefit “digital firms [that] already enjoy strong accumulated advantages. For example, platforms may have substantial market power at multiple levels related to the AI stack. This can give these firms the ability to protect against AI-driven disruption, or harness it to their particular advantage”.
Arrangements involving key players that could amplify risks. “Partnerships, financial investments, and other connections between firms related to the development of generative AI have been widespread to date.” These “could be used by major firms to undermine or coopt competitive threats and steer market outcomes in their favor at the expense of the public.”
The Joint Statement also stresses that “common principles to enable competition and foster innovation” in AI are “fair dealing,” “interoperability” of products and services across platforms, and “choices among diverse products and business models” for businesses and consumers.
But there’s a big problem. The theoretical concerns about the manipulation of AI to promote “control of key inputs,” “entrenchment,” and “anticompetitive arrangements” are at odds with reality.
Professor Jonathan Barnett of the University of Southern California recently addressed the concerns the Statement raises in a detailed factual analysis of the state of AI competition:
“Relying on the assumption that digital platform markets are prone to converge on entrenched monopoly outcomes, some commentators and regulators favor intervening preemptively in the [AI] ecosystem. This contribution assesses whether there are reasonable antitrust grounds for taking such action. Available evidence indicates that technically competitive entrants can generally secure access to the inputs required to achieve entry, including funding, semiconductors, cloud-computing services, datasets, and foundation models. Consistent with this view, entry into the models and applications segments of the [AI] ecosystem is especially robust. Investments and alliances involving large technology platforms, venture-capital, and institutional investors, and model developers, which have elicited regulatory concern, currently appear to be efficient arrangements to aggregate the complementary assets required to produce [AI] models and applications and face competition from other business models.”
The Joint Statement is not only wrong on the facts. Its “common principles” that should guide enforcement are also flawed.
“Fair dealing” may sound good, but it is a subjective standard that may be applied to benefit companies favored by government at the expense of others that are more efficient. In reality, vigorous competition on the merits that benefits the economy ultimately is the “fairest” outcome for all.
Mandating “interoperability” among high tech platforms raises multiple problems. It may promote collusion among firms (the prime evil of antitrust), and it is hard to enforce, requiring error-ridden oversight by courts. This oversight is costly and time-consuming. Interoperability also imposes expensive design requirements on high tech platforms, allowing third party “free riders” to gain access. This reduces platforms’ incentives to invest in service improvements, to the detriment of platforms’ clients. Dynamic competition is reduced, not enhanced.
Promoting “choice” in isolation is a meaningless concept. Competitive markets in general deliver a wide variety of new offerings for consumers. Allowing government to decide what and how many “choices” are appropriate involves second-guessing market outcomes, a role for which government is ill-equipped.
The Joint Statement Does Not Factor In AI’s Benefits
The Joint Statement spins its wheels contemplating imagined AI-related competitive harms, while ignoring the economic benefits bestowed by AI.
The potential economic gains from AI are enormous. A 2023 analysis by Goldman Sachs found that AI “could drive a 7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period.”
Studies show that specific AI benefits include increased worker productivity, innovation across the economy (with major gains in specific fields, such as health care), and scientific advancement.
Indeed, “[a] study by Accenture found that artificial intelligence could add $14 trillion to the global economy by 2035 with the most significant gains in China and North America. The study also predicted that AI could increase labor productivity by up to 40% in some industries.”
Moreover, analysis indicates that by “leverag[ing] vast amounts of data and predict outcomes. AI can significantly improve decision-making. It can optimize production, enhance product quality, and reduce waste.”
The DOJ And the FTC Should Disavow the Joint Statement
The Joint Statement casts a pall on AI without justification. As such, it may be expected to detract from aggressive, welfare-enhancing investments in AI improvements. It may also curtail innovative joint arrangements among digital platforms and small AI innovators (of which there are many), thus diminishing innovation-driven competition. U.S. AI leadership may be threatened. The losers will be American consumers and businesses.
The FTC and the DOJ therefore should publicly disavow the Joint Statement, which reflects the ill-considered precautionary and regulatory mindset of European competition enforcers.
The U.S. leads in high tech innovation, while the Europeans are laggards. The last thing the U.S. should do is emulate European high tech competition policy.
