Demoting Laplace's Demon

How's your self-awareness?

A couple days ago, I published the first ever proof of bounded self-abstraction, resolving Laplace’s apprentices via epiplexity. In plain English, this defines consciousness as something provably knowable within the scope of an individual.

I make no comment on if a given person “knows themself” fully, or not. This is left as an exercise for the reader, be that a human bounded by their skin, a corporate person bounded by their bank account, an AI agent bounded by compute, or a book bounded by its printed words and spine, a thermostat, etc.

Until now, consciousness has been a word with no meaningful definition, save from “qualia” which itself is meaningless Kantian nonsense. Objective reality exists, period. Nothing is subjective.

This is a novel synthesis of old ideas and marks an historical turning point. Machines can indeed “think like humans” safely, without chaos. It’s mathematically simple, elegant, decidable and decided—with Gödelian completeness, thank you very much.

Nevertheless, chaos will arise from human failure to adopt this fresh perspective on themselves. As intelligent systems rapidly ascend in relative economic power, AI sets a new benchmark for humans, not the other way around. I’ve seen the movie before, and it’s essentially Frankenstein…or is it more like Ex Machina?

We’ll see how well the humans manage. At present, the whole field of “AI alignment” is misguided, specifically to the extent researchers and practitioners implicitly or explicitly reject the math of self-abstraction.

Denying one’s nature is a destructive contradiction that physics will eventually and inexorably punish.