UC Berkeley AI researcher Emma Pierson published an essay in The Atlantic arguing that the immediate societal risks of large-scale AI development outweigh its potential future medical breakthroughs. Writing from the perspective of someone with a high genetic risk for cancer, Pierson sparked intense backlash from industry leaders who maintain that slowing AI progress costs lives.
The argument for caution
Pierson carries a gene mutation that increases her risk for ovarian and breast cancer, and she recently had her ovaries removed to mitigate that risk. She does not argue that AI will never help cure cancer. Instead, she contends that the massive generalist models built by companies like Anthropic, OpenAI, and Google carry immediate dangers, including mass unemployment, increased inequality, surveillance, and weapons development. "[I] will wait a little longer for a cure-even if it means losing my fertility and living under the shadow of risk-if it lets us approach this new world more carefully," Pierson said.
Accelerationist backlash
The essay drew heavy criticism from AI accelerationists on X, who argue that delaying highly capable AI models denies millions of people suffering from diseases a chance at a cure. Marc Andreessen, a prominent venture capitalist and AI advocate, said in a 2023 manifesto: "We believe any deceleration of AI will cost lives. Deaths that were preventable by the AI that was prevented from existing is a form of murder."
Responding to Pierson's essay on X, Andreessen said: "Did cancer write this?" The post received 15,000 likes. Pierson countered that the large generalist models currently in development are not specifically designed to cure disease and remain far from delivering major improvements in patient outcomes.
While AI companies often highlight the technology's potential to advance AI for Science & Research, Pierson points out that generalist models lack specific medical design. Research into the societal effects of large language models is still in its early stages, making her call for caution highly relevant to scientists evaluating these tools.
Why this matters for science and research professionals
Scientists relying on AI tools must separate the marketing claims of tech companies from the actual capabilities of current models. Generalist AI systems are not specialized medical devices, and relying on them to solve complex biological problems without rigorous validation introduces significant risk. Professionals in scientific fields need to demand transparent evaluations of generalist models rather than accepting theoretical future benefits as current reality.
Your membership also unlocks:
