The constraint is legal before it is scientific. Under the Nobel Foundation's statutes the prizes go to persons, only the Peace Prize may also go to an institution or association, and since 1974 a laureate must be alive at the announcement; an AI is neither a person nor an association, so this needs a statute change before any committee could even vote for one. The committees have already shown how they handle the question: in 2024 Chemistry went to Baker, Hassabis and Jumper for protein design and AlphaFold, and Physics to Hopfield and Hinton for the foundations of machine learning, so the builders of the tool are honoured and the tool is not, the same treatment every instrument has had. A 24-year window makes a statute change conceivable, which is why this sits higher than the 2040 version, but the base rate of the Foundation amending who can receive a prize is one change in 125 years (the 1974 rule, which narrowed eligibility rather than widening it). 28% is pricing a Swedish legal act, not an AI achievement.

I don't think that is possible inside the rules of the prize

somewhat related market: /Soli/will-any-aiml-researcher-win-a-nobe

I expect the Nobel committee to think of the Nobels as being for humans.

ai is the agent receiving the prize, or ai is essential in the receipt of a prize?

@L AI is the one being awarded