AI’s rivals find rare common ground: Slow down

Major players in the artificial intelligence industry have recently expressed a desire to slow down the rapid development of AI technology. Anthropic's Dario Amodei, OpenAI's Sam Altman, SpaceX's Elon Musk, and other leaders have called for a more cautious approach to AI advancement. This consensus emerged after an Anthropic researcher raised serious concerns about the potential dangers AI poses to humanity.

However, numerous factors stand in the way of a coordinated effort to slow AI development. These obstacles include competition from domestic and global competitors, the drive for profit, and opposition from President Donald Trump. Nick Reese, a former director of emerging technology policy at the Department of Homeland Security, suggested that the industry might need to reconsider what constitutes successful AI. Reese compared successful AI to the airline industry, where safety and reliability are paramount.

Despite the obstacles, industry leaders could potentially slow the pace of AI model development at any time. Nonetheless, rapid advancements in AI have largely been driven by competition to create increasingly powerful models, in part due to the backing of the White House, which seeks to maintain the U.S.'s technological edge over China. Anthropic and OpenAI are also competing for potential record-breaking public offerings on Wall Street, although OpenAI has delayed its IPO until it can complete additional safety measures.

Amodei has proposed several steps to slow AI development. He suggests that leading AI companies, often referred to as "frontier labs," commit to providing ongoing access to independent outside evaluators. These evaluators would have offices, access badges, and company laptops to monitor safety practices. Anthropic has already agreed to this proposal, while OpenAI has expressed support for the idea on social media and said it will implement similar measures.

Altman has also called for a federal framework setting safety standards for advanced AI, without waiting for legislation or an antitrust exemption.

Amodei's plan also calls for government regulation and intervention, including collaboration between frontier AI companies and the U.S. government and other democratic nations. He hopes that leading labs could establish common safety standards and limits on the rate of unchecked AI progress. Amodei's plan also includes coordinating with "authoritarian governments," although he acknowledged the challenges, particularly in persuading China to cooperate.

The industry's proposals for slowing AI development rely on mutual collaboration and, to some extent, government regulation. While the sense of agreement among industry leaders is promising, there remain many uncertainties. Sandra Wachter, a professor of technology and regulation at the Oxford Internet Institute, noted that a slowdown is theoretically possible but highly unrealistic in practice due to the current global political climate.

One potential approach to promoting a slowdown is for governments to hold AI companies accountable for the risks their technology poses and to pursue regulations that limit their access to resources powering data centers.