Exercise physiologist Mike Israetel would accept a 49% risk of human extinction for a chance to live without aging and become more intelligent in a thought experiment
Exercise physiologist Mike Israetel would accept a 49% risk of human extinction for a chance to live without aging and become more intelligent in a thought experiment
In a debate released on November 11 on The Beyond Tomorrow Podcast, Israetel and AI safety researcher Roman Yampolskiy disagree over how to extend human life. Israetel places his hopes in autonomous superintelligence, an AI that outperforms humans at every intellectual task. Yampolskiy proposes using AI for specific tasks set by humans.
Israetel puts the actual probability of human extinction caused by superintelligence at almost zero. Yampolskiy proposes a thought experiment: suppose the probability of catastrophe were known exactly. Under those conditions, Israetel would accept a 49% risk of human extinction for a 51% chance of success, because he believes humanity will almost inevitably die out if it remains in its current form. For him, success means becoming a more advanced form of life.
Israetel believes that a superintelligence would benefit from cooperating with humans. In his scenario, the machine would initially need human labor and would later need to study the complexity of the living world to test its models against reality. By destroying humans, it would lose valuable data. Yampolskiy responds with the example of animal experiments: an interest in studying living beings can coexist with causing their suffering and death, and a few specimens may be enough for experiments. He also considers the need for human labor temporary, lasting only until robots can replace people.
Yampolskiy proposes directing AI toward specific tasks: discovering drugs, treating diseases, and extending life. As Yampolskiy puts it, “humans decide what to do, how, and why.” Israetel believes human oversight would limit scientific inquiry because people can ask only the questions they can conceive of. He envisions an autonomous superintelligence that could formulate and solve problems beyond human imagination. He is therefore willing to let the machine choose the direction of research as well.
The host asks Yampolskiy to imagine being revived after having his body frozen, at the cost of a 50-percent risk of human extinction, including the deaths of Yampolskiy's children. Yampolskiy refuses. Israetel agrees that saving oneself this way would be selfish if the risk of human extinction were far above 50%, as Yampolskiy believes it is. Israetel wants everyone to be safe and anyone who wishes to be able to live free of aging and disease and become more intelligent.
When discussing a digital copy of the brain, Israetel describes a scenario in which he is still alive in his biological body while his digital version speaks to him from a screen. He is content for the copy to remain after the original dies. Yampolskiy also prefers preserving a copy to losing all information about himself, but he wants his own existence to continue. Faced with a choice between death and digital life, he says, “I would choose existence.”