Project Q* will not be subject to hallucinations, but epiphanies

The technology behind the most chimerical project of the moment is a paradigm shift compared to the GPT transformer. And it could change the rules of the game. Brief guide to avoid replacing the distortions of long-termism with those of short-termism.

Yes, it is likely that behind the embarrassing theatrics staged by the leaders of OpenAI, which discredited what is one of the most visible companies in the world, there really is this mysterious “Q* project”. Reuters reported it first, then many joined in, but the blanket of fog around the project itself has not dissipated. On the contrary.
What do we know about Q*, then? At the moment, just a dense web of assumptions, which have a single common thesis: it seems that Q* is good at doing mathematical operations. It is not known how complex it is — and knowing it would substantially change our level of anxiety — but ultimately this is enough to understand that, if he really knows how to do them, and not just estimate the statistically most probable number as a solution, this object is very different from what we’ve seen so far in the most recent wave of generative AI that newspapers have started to cover.
The problem is…