Live·Open questions in longevity research
All news
Industry & policy

Ruxandra Teslo estimates that 30–40% of AI safety professionals outside AI labs accept some form of handing control of humanity’s future to AI

12 November 2026· 261112005

Ruxandra Teslo estimates that 30–40% of AI safety professionals outside AI labs accept some form of handing control of humanity’s future to AI

On November 11, Ruxandra Teslo published an essay titled “Handing Over Humanity.” Teslo, a contributor to Works in Progress, describes and criticizes “handoverism”: humanity survives, while AI determines its future.

The essay opens with a developer at a major AI lab describing humanity’s permanent loss of control: everyone would be ruled by a hopefully benevolent AI god. “This is not my preferred outcome, but once we build a superintelligence that can no longer be stopped, I see no other path for humanity.” He added: “It is still much better than what I actually fear most: that we might all die.”

Teslo distinguishes handoverism from successionism, which accepts the replacement of humans by AI. Under handoverism, the ruling AI’s goals are aligned with human goals. Teslo takes as her starting point an essay by blogger Scott Alexander, who defends handover and describes five versions of it. The first begins with a concern: democracy is protected by the ability of workers and soldiers to rise up against a tyrant. AI will probably replace them, allowing a small group of people, or someone such as a president, to seize power. The fifth is called “God”:

“If we do build a godlike AI and decide, to our own satisfaction, that it is fully aligned, I would prefer to let it choose its own direction and humanity’s direction rather than leave it under the control of the US president, a shareholder vote, or something similar.”

Based on her conversations, Teslo estimates that 30–40% of the AI safety community outside AI labs accept handover, as she writes on X. She puts the proportion higher among lab employees in core AI roles. Some see rule by an AI that surpasses humans as a good future; in Teslo’s reading, Alexander belongs to this group. Others have resigned themselves to handover because they consider it inevitable or preferable to extinction, without necessarily wanting it. Teslo’s conversations suggest that this view is particularly common outside the labs.

Almost every conversation about a future with human-level AI promises an end to disease. Teslo’s sense is that this need not require a godlike AI: medicine is already moving toward substantial progress through experimental science and existing AI models. She adds that any assessment of progress must consider the kind of society it will produce. “After all, why do we want to cure disease, if not to live a good life for longer? Take away the good, and simply staying alive starts to seem less appealing.”

Teslo’s first argument against handover concerns self-determination: people’s ability to manage their own affairs. The thinker Tocqueville warned of a gentle form of power that would relieve citizens of the need to think and the burdens of life. Another argument is practical: perpetual AI rule would require aligning its goals with human goals once and for all. Teslo doubts that this would work, because government responsive to people has required institutions built over centuries through feedback and error correction. A further concern is permanence: “Why should the beliefs of 2026 become the permanent constitution of humanity’s future?”

Alexander qualifies his position in his essay, saying that the values embedded in AI “should include the understanding that we can debate our values and ultimately change them.”

Teslo sees the future as still open: on the threshold of recursive self-improvement, when each generation of AI begins creating the next, people still have a choice. That choice needs prompt and broad discussion.

Originally published on Telegram by Ukhvat NewsView on Telegram
Sources
#ai-safety#handoverism#superintelligence#ai-governance#human-self-determination