Superintelligence Should Be Built Disabled, Position Paper Argues

A position statement published in AI Perspectives & Advances argues that the dominant model of artificial intelligence development—measuring progress by accumulating capabilities such as more parameters, more tasks and more autonomy—is fundamentally misguided. The paper, authored by He Huang and published open access on 15 September 2026, proposes instead that beneficial superintelligence should be pursued by deliberately engineering limitation rather than maximizing ability.

The argument draws on three core insights from disability studies: that limitation is a fundamental condition of existence, that subjects are interdependent, and that ability is relational and contextual rather than absolute. From these premises, the author contends that a truly super-AI would be defined not by what it can do, but by its profound understanding of what it cannot do, together with its capacity to communicate those limitations clearly to human users.

The framework also offers an alternative to what the paper calls the perilous fantasy of autonomous super-agents—systems granted maximal capability and independence—by grounding superintelligence in acknowledged interdependence with humans rather than in unchecked autonomy.

The paper further presents this relational view as a basis for evaluating AI capability. As a position statement, the article sets out a conceptual and philosophical proposal rather than reporting experimental results; it does not describe a implemented system or empirical tests of the approach. Its claims about how engineering meaningful disability would perform in practice therefore remain arguments to be developed and tested by future work. The article is currently available as an early-access, peer-reviewed accepted version, citable under a permanent DOI, subject to further edits before the final Version of Record appears.

Subject of Research: A philosophical position paper proposing disability as a design principle for beneficial superintelligence.

Article Title: The disabled AI is the super AI: disability as a design principle for superintelligence

Article References: Huang, H. (2026). The disabled AI is the super AI: disability as a design principle for superintelligence. AI Perspectives & Advances. https://doi.org/10.1186/s42467-026-00019-4

Image Credits: AI Generated

DOI: 10.1186/s42467-026-00019-4

Keywords: artificial intelligence, superintelligence, disability studies, AI hallucination, autonomous agents, AI capability, position paper, AI safety, relational ability, AI design, disabled, super