In the 1960s, computer scientist Ray Solomonoff introduced a universal approach to knowledge, based on a combination of probability theory and theoretical computer science. This approach can be argued to be the ultimate solution to artificial general intelligence. Indeed, Solomonoff proved that it was complete. Unfortunately though, it is also uncomputable. Intriguingly, Solomonoff also proved that completeness and computability were incompatible properties. The talk will be revisiting Solomonoff’s reasoning.
Bienvenue à tous!