The public is five times likelier than AI researchers to believe AI is conscious: attachment and honest design norms

The people most likely to believe AI is conscious are not the people who build it.

Byrne Hobart, writing in The Diff, cites the last survey he has seen on the question. It found that belief that AI models are conscious was about five times more prevalent among the general population than among AI researchers, although the researchers were more open to the idea that computers could eventually be conscious. That cuts against the popular story that Silicon Valley is pushing the idea onto a sceptical public.

He also points to a New York Times report by Elizabeth Dias that Anthropic nearly pulled out of consulting with the Vatican on its encyclical, Magnifica Humanitas, over concern it would declare AIs non-conscious. Hobart thinks the row wildly exaggerates both sides' confidence. He quotes Chris Olah of Anthropic: "we don't know if A.I. models are conscious. I don't know. I'm genuinely uncertain." And he quotes Cardinal Michael Czerny calling the question serious and one that deserves further study.

I am not equipped to settle the philosophy, and neither is a survey. For business the behavioural question matters more. Customers form attachments to systems they experience as someone, whatever the metaphysics. That shapes trust, loyalty, grief when a product is withdrawn, and the expectations people bring to a chatbot that sounds like it cares.

It also creates a design obligation. A company can encourage that perception or temper it, and the commercial incentive usually points one way. Regulators are more likely to arrive at this through consumer protection than through the hard problem of consciousness.

The firms that set honest norms now will not have to explain later why their product encouraged a belief they could not support.

Source: Byrne Hobart, The Diff, "Longreads + Open Thread" (3 October 2026), https://lnkd.in/g-weKKtA

#AIEthics #ConsumerBehaviour #ResponsibleAI