Will superintelligence gain linear utility from just thinking longer?
enormochungus · x · 2026-11-19
The author floats a possibility: superintelligences may be so much better at thinking that the utility of additional thinking on a problem is linear. He finds this absurd and would bet against it, but admits he can't give a good reason why. The replied context suggests it may be false because superintelligences could have far better ways of recursively defining Turing machines, reconstructing informal parsimony in efficient practice.