Bedoya: FTC Was Focused on AI Competition Before Biden Order

President Joe Biden’s executive order on AI (see 2310300056) didn’t include FTC recommendations the agency wasn’t already exploring, Commissioner Alvaro Bedoya said Wednesday.

Speaking at an Open Markets Institute event, Bedoya said he was already focused on issues related to competition, data dominance and bias, all laid out in the EO. Biden’s order suggested the agency use its rulemaking authority to ensure fair competition in the AI marketplace protects workers and consumers. Bedoya said he's particularly focused on the labor-related aspects of AI technology.

Bedoya said AI technology can create barriers for market entrants and further entrench dominant platforms. Vertically integrated companies have a major advantage given the ability to feed their leading AI models with large amounts of data and pair those models with ubiquitous platforms, he said. Bedoya didn’t name specific companies but said, “I’m sure every single person in the room can think about which market actors have these” foundation models, data and platforms.

Tech companies have used and will continue to use competition with China as a justification for pushing back against regulation, said Rana Foroohar, Financial Times global business columnist. The China competition angle is a “disingenuous argument” that distracts from the fact that these tech companies are “eating everyone’s economic lunch,” she said.

However, the U.S. is in a “real” economic contest with China, said Tim Wu, Biden’s former special assistant on tech and competition policy. “I do take the idea of an arms race seriously, and I take the idea of it mattering seriously,” said Wu, now a Columbia Law School professor. But these “national champion” attitudes toward regulation historically end in arms race losses, he said. The better approach is to promote domestic competition through fair and open markets that allow invention and entrepreneurship, said Wu.

Northeastern University Assistant Professor Laura Edelson agreed with his argument against U.S. favoritism. AT&T’s dominance in the 1970s and 1980s is just one example of a domestic monopoly that hampered U.S. innovation, she said. Regulators should also be wary of the amount of control tech companies have over personal data, she said. Wu urged policymakers not to rush regulation, given the many open questions on AI. The government should tailor policies to actual, concrete AI harms, not perceived risks, said Wu.

Bedoya said competition problems exacerbate issues related to technological bias. This includes automated decisions related to employment, healthcare and insurance claims. “People don’t appreciate how pervasive these systems are in the major decisions around us,” he said. “I will be asking the most [enforcement-related] questions about” these systems.

Bedoya credited Chair Lina Khan for reexamining FTC Act Section 5 and finding ways to ensure the agency is honoring the statute’s original intent. Section 5 allows the FTC to track good and bad innovation, including situations where dominant market participants are using their power to remove rivals. Section 5 authority will be key for the agency to address AI-related labor claims in the entertainment industry, he said.