Ariel Zeleznikow-Johnston proposes preserving the brain after death to reduce incentives for risky AI acceleration
Ariel Zeleznikow-Johnston proposes preserving the brain after death to reduce incentives for risky AI acceleration
On August 27, neuroscientist Ariel Zeleznikow-Johnston published an essay in the Brain Preservation Foundation’s official newsletter about biostasis, the preservation of a dying person’s brain using chemicals and cold. He argues that this procedure could give people time to await future medical advances and make them less willing to accept risks in pursuit of rapid AI development.
Zeleznikow-Johnston begins with a dilemma. Superintelligence, an AI system capable of surpassing humans in intellectual tasks, could accelerate the search for new treatments. In his model, losing control of such a system could be catastrophic.
In Nick Bostrom’s calculation of the cost of a pause, deaths during the waiting period make acceleration especially attractive to people who need future medicine within their lifetimes. David Wood proposed slowing this race while investing more heavily in biology.
In his new essay, Zeleznikow-Johnston applies this calculation to individual decisions. Biostasis uses chemicals and cold to stabilize a dying person’s brain. The author links long-term memory and personality traits to the brain’s physical structure. He argues that preserving this structure may give the person a chance to await future methods of treatment or neural information retrieval.
“Make death less likely, and taking major risks will no longer seem as reasonable. If you knew that you would still be here in fifty years, regardless of whether or when superintelligence arrived, you would have far fewer reasons to bet on dangerous AI acceleration.”
In the author’s model, biostasis gives a person another way to await future medicine, which reduces the expected benefit of risky AI acceleration.
Zeleznikow-Johnston proposes testing the quality of brain preservation, making the procedure more widely available, and reducing its cost. He argues that a more reliable and accessible reserve of time would reduce the personal urgency that leads people to support risky AI acceleration.