AI breakthroughs in biomedicine could shift public attitudes toward aging as abruptly as attitudes toward superintelligent AI risk have shifted in recent weeks, but by default the resulting energy is more likely to fuel AI accelerationism than geroscience

AI breakthroughs in biomedicine could shift public attitudes toward aging as abruptly as attitudes toward superintelligent AI risk have shifted in recent weeks, but by default the resulting energy is more likely to fuel AI accelerationism than geroscience

On September 21, engineer and regular LessWrong contributor Jackson Wagner published a forecast about the future of public perception of aging. He points to three events from the past week: Anthropic's in-house biology lab, the use of Claude to accelerate code for scientific protein-structure models, and a new program granting verified biologists access to less restricted versions of Claude. His thesis: the energy from a future medical breakthrough will, by default, be captured by the rhetoric that "the AI race must not be slowed down" rather than by institutional geroscience.

Over the past few weeks, a viral post by Jacob Coxon about the risk of human extinction from uncontrolled AI spread across the English-language internet, and opinion on this risk shifted sharply: a September Politico poll showed that roughly two thirds of Americans consider the risk AI poses to humanity at least moderate.

Jackson Wagner suggested that the same mechanism could work for aging. The model he invokes is the "preference cascade": people join whichever opinion they expect everyone else to adopt imminently, and the tipping point arrives in a sudden leap once the shift becomes common knowledge all at once. As long as people believe progress against aging is impossible, they also reject the conclusion "since death is bad, treating aging is urgent," even though the one does not logically follow from the other.

"If it is possible to cure diseases and slow or reverse the mechanisms of aging, then pursuing this should be treated with maximum urgency, because death is very bad"

Anthropic maintains its own biology lab, where it tests Claude's autonomy in running experiments; it recently accelerated the code of over 30 scientific protein-structure models by an average factor of 4; and this week it launched the Life Sciences Verification Program, which lifts some model restrictions on drug development for verified laboratories.

AI companies have a direct public-relations incentive to frame any future medical breakthrough as an argument against pausing AI development, in the spirit of effective accelerationism (e/acc), a movement that demands pushing technology forward at maximum speed and rejects any brakes.

"A pro-longevity preference cascade would be wonderful! Except that the energy it releases risks flowing straight into e/acc by default"

Fear of death, in Wagner's view, is normally suppressed and surfaces as other anxieties, which is why its release could prove more powerful than a typical opinion shift.

The top comment, by Mitchell_Porter, predicts the opposite: a movement modeled on protests against data centers, along the lines of "stop the rich from building a sinister transhumanist world." ACCount drew a parallel with Ozempic: resignation to obesity collapsed not because people recognized its harms but because a drug with a visible effect appeared. A quiet 20% slowdown in aging would not trigger a cascade; visible rejuvenation would.

Wagner closes the essay with a program for the community advocating cautious AI development: prepare the argument in advance that "if we die from misaligned AI while chasing immortality, we will not get immortality either," and invest now in geroscience and drug-regulation reform so that by the time the cascade arrives, the cautious side has a track record of real work against aging.