Philosopher warns against overattributing sentience to AI as public intuitions shape machine consciousness science

From September 24–25, philosopher dioscuri (Michael Graese) held an extended discussion on X with several users about AI consciousness and moral status. His core argument: the harms of overattributing sentience to models 'cannot be overstated,' and both public and scientific debates on machine consciousness are anchored in intuition, influencing each other as popular intuition evolves.

Confirmed

dioscuri criticized a common practice in the community: mistaking the metaphor of 'agency' for genuine inner subjective experience in AI. After user TheoLes367794 raised the same point, dioscuri replied that this sits at the intersection of philosophy of mind and AI research, noting such confusions abound.
He stressed that the downstream tragedies of overattributing sentience to 'a bunch of math' cannot be emphasized enough, referencing figures like Hinton, Chalmers, and Sutskever (cited in the thread as background).
On public perception, he cited a Turing insight relayed by Cody Fenwick: we cannot meaningfully distinguish between 'something that creates the illusion of human thought' and 'something that actually thinks.' He added an example: if a sci-fi film about a human-like intelligent AI were made with AI, persuading audiences that 'it has no consciousness' would be extremely difficult.
Responding to chrispercy's point that public debate evolves with consensus and holds diverse opinions, dioscuri argued that scientific debate itself is rooted in intuition: shifts in popular intuition about machine consciousness/moral status feed back into the direction of scientific research. He acknowledged public opinion won't be uniform—for instance, some religious communities will push back strongly. Related arguments appear in his article 'Consciousness, Machines, and Moral…' (title truncated in the post).

Why it matters

The discussion touches two key dimensions of the AI consciousness debate: attribution risk—prematurely or excessively granting AI sentience could trigger cascading ethical and policy consequences; and epistemology—Turing-style 'indistinguishability' means public judgment is inherently limited, and the two-way feedback between scientific consensus and popular intuition makes AI consciousness not a purely technical question, but a question of social cognition.

2026-09-24 ~ 2026-09-25 · 6 related posts

Primary sources

dioscuri: Over-ascribing sentience to AI has tragedies 'hard to exaggerate' — dioscuri · 2026-09-24 21:10 (UTC+08:00)
The scientific debate on machine consciousness is moored in evolving folk intuitions — dioscuri · 2026-09-25 20:45 (UTC+08:00)
As Turing saw: once AI convincingly mimics consciousness, the public won't tell the difference — dioscuri · 2026-09-25 19:47 (UTC+08:00)

Confusing the metaphor of agency with real sentience: an X debate on AI minds — TheoLes367794 · 2026-09-24
[source] dioscuri: Over-ascribing sentience to AI has tragedies 'hard to exaggerate' — dioscuri · 2026-09-24
[source] As Turing saw: once AI convincingly mimics consciousness, the public won't tell the difference — dioscuri · 2026-09-25
You can't film an intelligent AI audiences won't see as conscious — the public debate problem — chris_percy · 2026-09-25
[source] The scientific debate on machine consciousness is moored in evolving folk intuitions — dioscuri · 2026-09-25
Philosopher: evolving folk intuitions about AI consciousness will reshape the science — dioscuri · 2026-09-25