The Last Invention

3 min read

Radical life extension

Ending aging as a cause of death, so that people live for centuries in good health rather than decades. For most of history this was a fantasy. Advanced AI has made it a serious possibility within the lifetimes of people alive today, and that changes how some of them weigh the risks of building it.

Why AI makes it plausible

Aging is a biological process, and biology is a field where AI may speed up research the most. Dario Amodei’s Machines of Loving Grace (2024) argues that AI could compress a century of biological progress into a decade or so, and that doubling the human lifespan to around 150 would only continue the trend of the 20th century, when life expectancy roughly doubled. Past that point, people could live long enough for further advances to keep extending their lives indefinitely: “longevity escape velocity”. Whether biology allows it is unknown, and Amodei says there is no guarantee.

How it changes the calculus

  • Delay has a death toll. About 170,000 people die every day, most of them from aging and disease. If superintelligence could end that, every year of delay means tens of millions of deaths that a working cure could have prevented. This is the strongest version of the “benefits are enormous” case in Why people build AGI despite the danger.
  • It makes large risks look acceptable, to those alive now. If the prize is centuries of healthy life, a gamble with a high chance of disaster can look worth taking. Optimal Timing for Superintelligence works this out for people alive today and finds even very high risks acceptable, a conclusion critics call selfish.
  • Extinction is not just death. The main reply: people dying of old age leave descendants, while extinction ends the human story. Eliezer Yudkowsky puts it bluntly: anyone who wants to live forever should sign up for cryonics rather than gamble with “your neighbors’ children’s lives”. Trading a high chance of extinction for longer lives today counts only the present generation, and no one asked the people who would bear the risk; see Consent.
  • It is a personal motive. Some people building AI, and some cheering it on, want to live long themselves. A personal stake of this size makes the risks easier to discount, whether or not anyone says so.

Progress without AI changes the picture

The pull towards building fast depends on aging being unsolved without AI. If medicine made progress against aging by other means, the cost of waiting would fall, and a longer, safer path to superintelligence would become easier to accept, since most people alive could expect to live to see it. In that sense, research on aging is also, indirectly, a way of reducing the pressure to race; see When should AI development pause.

What people would do with very long lives in a world where AI does the useful work is the question of Deep Utopia.

Backlinks