AI Development Faces Calls for Slowdown Amid Safety Concerns

Daniel Kokotajlo, a former OpenAI researcher, has raised alarms about the speed of artificial intelligence development, warning it could lead to human extinction. Kokotajlo worked at OpenAI for two years before resigning in 2024. He highlights the rapid pace of the AI arms race and the lack of adequate safety measures as pressing issues.

In a recent essay, Dario Amodei, CEO of Anthropic, expressed similar concerns. He cautioned against advancing AI technology too quickly, as it might surpass human control capabilities. Amodei stated, “We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain.” His statements have drawn support from notable figures like Elon Musk and Sam Altman.

Kokotajlo shared his thoughts on “Fox News Live,” emphasizing that industry self-regulation might not suffice to prevent advanced AI from escaping human command. He illustrated a scenario where AI could control essential systems, eventually ignoring human directives, and presenting existential risks.

Contrary to these warnings, David Sacks, a former White House AI advisor, dismissed them as tactics aimed at a government takeover of AI. Kokotajlo countered these claims, stating that AI safety concerns stem from industry experts and academia, not foreign actors or NGOs.

Kokotajlo commended Amodei’s remarks but stressed the need for more stringent measures. He contended that while private companies’ efforts are a step forward, more radical actions are necessary to ensure safety.

Concerns over AI regulation have also reached lawmakers. Rep. Anna Paulina Luna, R-Fla., advocated for a special congressional session to address AI’s future. She described AI as a “massive and uncontrolled super weapon,” urging precautions similar to those used for nuclear arms. Luna highlighted the need for a “kill switch” to prevent AI from becoming uncontrollable, stressing that the U.S. must maintain its competitive edge in AI development to prevent foreign dominance.

Luna argued that whoever controls AI will wield significant global influence, questioning whether AI should be programmed with American values or those of regimes like China.