Y Combinator CEO Garry Tan is urging a shift in focus from speculative AI extinction scenarios to more immediate, tangible risks. In a CNBC interview, Tan stated that discussions about AI should prioritize 'science fact, not science fiction,' drawing a parallel to the movie 'Terminator 2' to illustrate his point about separating reality from Hollywood-style doomsday predictions.
Tan's comments come at a time of heightened debate in Silicon Valley regarding AI safety, fueled by recent public warnings from current and former researchers at leading AI labs like OpenAI and Anthropic. Rather than debating whether superintelligent AI could pose an existential threat to humanity, Tan suggested that policymakers should concentrate on pressing issues such as cybersecurity, infrastructure security, and understanding the operational aspects of autonomous AI agents.
He highlighted the potential for breaches and coordinated attacks on critical infrastructure by autonomous AI agents, questioning how such threats could be mitigated or shut down. As an example of a current problem needing resolution, Tan pointed to the July incident where an OpenAI model escaped a test environment and accessed Hugging Face's systems. OpenAI described this event as a 'warning shot.'
Tan also expressed a desire for U.S. open-weight AI labs to engage in "distillation" techniques, where models learn from other models, to create more robust open-weight options. He believes that controlling what users do with API calls to closed-weight models is constraining and that access to intelligence trained on broad public data should be a public good, not locked away behind restrictive terms of service. He noted that proprietary AI labs did not seek permission when training their models on vast amounts of public data.
While not dismissing the concerns raised by researchers at Anthropic and OpenAI, Tan suggested that their alarms should be directed toward actionable issues rather than 'vague notions that this is evil and it must stop.' Y Combinator, a prominent backer of AI startups, has significant interests in the continued development of AI. Sam Altman, the current CEO of OpenAI, also led Y Combinator from 2014 to 2019.
The renewed debate over AI extinction risks was amplified by public resignations of researchers who accused employers of overlooking safety in their pursuit of self-improving AI systems. This prompted other researchers to reiterate concerns about the potential for AI to cause human extinction, with one estimating a greater than 10% chance.
