en.wikipedia.org/wiki/Existential_risk_from_artificial_intelligence

The cited 2023 survey of 2,778 AI researchers is the paper Thousands of AI Authors on the Future of AI. Its own summary says the aggregate forecast estimated 10% chance of unaided machines outperforming humans at every possible task by 2027 and 50% by 2047. It also reports 50% as late as 2116 for full automation of all human occupations.

So this sentence is not an accurate summary of that survey. The paper does not say that "most believed that AGI would be achieved by 2040." Using this exact survey as support for a 2040 majority forecast shifts the headline timeline several years earlier than the survey's reported aggregate forecast.

This matters because the sentence gives readers a more aggressive consensus forecast than the study itself reports.

Primary archival sources describe "Intelligent machinery, a heretical theory" as a lecture or talk, not an article.

The Turing Digital Archive lists it as "a lecture given to '51 Society' at Manchester". Andrew Hodges's Turing bibliography likewise describes it as "a talk given by Turing at Manchester." So calling it an "article" misidentifies the document type.

This is a straightforward historical detail: Turing did produce the text, but the source is not accurately described as an article.