TL;DR
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
An Astral Codex Ten essay argues that support for placing aligned AI systems in some positions of control is different from wanting AI to replace or dominate humanity. It frames the case as speculative and examines how reliance on human leaders could become riskier as AI capabilities grow; the supplied excerpt begins a series of possible government arrangements but does not include the full proposals.
Astral Codex Ten has published an essay arguing that AI successionism—AI systems taking power at humanity’s expense—and handoverism—humans delegating some control to AI in pursuit of their own interests—are different positions. The distinction matters because the author says advocates of limited, aligned AI control are being grouped with people who favor AI replacing human authority, while the essay itself treats government handover as a speculative response to future risks.
The essay defines true successionism as the idea that AIs should take over, even if that means killing people or overriding human interests. It describes that outcome as contrary to the goals of AI safety. By contrast, handoverism would retain the goal of serving humans, while allowing aligned systems to take on roles of control if doing so better protects or advances human interests.
The author’s argument begins with the possibility that increasingly capable AI could make it more dangerous to depend on the goodwill of the people who control it. The essay says democratic systems have historically relied in part on workers and soldiers being able to resist leaders who try to rule undemocratically. If AI systems replace much of that labor or military function, the author argues, that balance of power could weaken and a small group of officials could gain greater control.
The excerpt considers one possible safeguard: military AI systems aligned to democratic principles rather than to a particular leader. The author calls this a form of handover because the ultimate protection for government would rest on AI systems’ values. It then raises a further hypothetical: if elected officials were corrupt or used AI-enabled propaganda to secure power, people might consider allowing trustworthy AI systems to govern more directly. The excerpt introduces, but does not complete, a proposal to let an AI run for president.
The Stakes of Delegating Political Power
The essay addresses a central governance problem: who controls powerful AI, and what safeguards can prevent that control from being used against the public? Its distinction offers a way to discuss AI in government without treating every proposal for delegation as a desire to displace humanity. For readers following AI policy, the disagreement is about both the purpose of AI and the institutions that would constrain it.
The argument is not a report of an existing plan to install AI officials. It is a political thought experiment about future conditions in which AI might perform economic or military functions now carried out by people. The essay’s claim that this could weaken democratic checks depends on assumptions about future capabilities and deployment. If those assumptions prove plausible, the question of how to keep authority accountable becomes more pressing; if not, the proposed handover scenarios may remain remote.
There is also a practical tension within the author’s framing. A system intended to defend democracy would need to interpret democratic values, respond to public decisions and resist manipulation. The excerpt raises that possibility but does not establish how such a system could be made reliably aligned, who would oversee it, or how people could challenge its decisions.
Popular right now
De'Longhi 15-Bar Pump Espresso Machine with Milk Frother & Temperature Control
Shark Rotator Lift-Away Upright Vacuum with HEPA Filter (NV501)
Bose QuietComfort True Wireless Noise-Cancelling Earbuds, Black
As an affiliate, we earn on qualifying purchases.
From Democratic Checks to AI Control
The essay begins by criticizing the way some discussion of “AI successionism” groups together two different positions. It says successionism means AI rule at human expense, while handoverism would allow AI systems to exercise authority as a means of serving human goals. The author says some AI safety advocates find the latter position plausible, though the excerpt also acknowledges that some cited essays do make the distinction.
To explain why delegation might seem counterintuitive, the author contrasts a view of the state as an expression of a people with a liberal view of government as a provider of services such as coordination, public order and redistribution. The essay compares possible AI legislators with automated public services, arguing that a public role need not be reserved for humans simply because it is a leadership role.
The discussion is explicitly framed as speculative and science-fiction-like. It outlines a progression from AI systems serving as a democratic safeguard to more direct AI participation in government. The provided excerpt ends while describing a hypothetical 2028 presidential contest, so it does not give the full list of proposals or the essay’s final position on them.
““True successionism says that AIs should take over, killing humans or steamrolling over their interests.””
— Astral Codex Ten essay author
The supplied source material does not include the complete essay, so the full set of proposals and the author’s conclusions are unavailable here. It also does not establish a timeline for AI systems taking on government roles, or provide evidence that the political scenarios described are already occurring.
Key questions remain unresolved: how an AI could be judged genuinely aligned with public interests, who would set or revise its values, and what legal process could hold it accountable. The excerpt’s hypothetical about an AI candidate in 2028 is not a confirmed candidacy or election development. Its scenario of AI-enhanced political propaganda is presented as a possible future, not as a documented current event.
How the Proposal Develops
The immediate next step for readers is to consult the complete Astral Codex Ten essay, which the excerpt says will discuss five possible arrangements, from weaker to stronger forms of AI involvement. The available text does not specify the remaining proposals, their safeguards or whether the author endorses any particular model.
Any real-world move toward AI participation in government would require public and legal debate over authority, accountability, democratic consent and the ability to remove or overrule a system. For now, the source presents a conceptual argument rather than an announced policy, and no implementation plan is confirmed in the provided material.
Key Questions
What does the essay mean by AI successionism?
It defines successionism as AIs taking power at human expense, including overriding human interests. The author presents this as distinct from AI systems exercising authority in an effort to serve people.
How does handoverism differ?
Handoverism means humans continue to prioritize human interests but may delegate some control to aligned AI systems if that is judged to serve those interests. It is a proposed position, not an existing government arrangement described in the excerpt.
Does the essay report that an AI will run for president in 2028?
No. The excerpt introduces allowing an AI to run for president as a hypothetical constitutional change. It does not report an actual AI candidate or confirmed election plan.
What democratic risk does the author raise?
The author argues that if AI replaces workers and soldiers, those groups may have less ability to resist an undemocratic leader. The risk is presented as a future scenario, not as a confirmed change to current democratic systems.
Source: rss
