In a recent interview, AI researcher Timnit Gebru pushed back against the industry's focus on existential risk, calling it a 'harmful distraction' from real problems. Gebru, who was fired from Google after co-authoring a paper on bias in large language models, argues that the narrative is self-serving. She points out that the same investors and founders who stand to profit from AI companies, such as Anthropic's backers, also fund the institutions that warn about AI destroying humanity. This, she says, is not a coincidence but a way to portray AI as superhuman—both capable of solving all problems and posing a unique threat—which justifies massive investment and regulatory exceptions.
Gebru also addressed criticism that her earlier research, which described large language models as 'stochastic parrots' that merely repeat training data, is no longer relevant. She argues that the underlying harms—such as bias and the perpetuation of inequality—remain pressing. The existential risk debate, in her view, draws attention away from concrete, current harms that affect people today. She compared the situation to the need for existing laws to apply to AI companies, as argued by former FTC chair Lina Khan.
Her remarks come amid a broader ideological split in the AI community, where some believe AI could destroy humanity, while others, like Gebru, focus on immediate societal impacts. She acknowledges that her stance has even aligned her with some right-wing critics of the effective altruism movement, but insists the real issue is who benefits from the existential risk narrative—and what it distracts us from.