I have repeatedly been challenged to “prove” that human-level artificial intelligence is impossible. My position is that the burden of proof lies on the opposite side.
No serious evidence has been given in support of the thesis that human-level AI is possible. There certainly is no reason to expect that it would be.
And there are very good reasons to believe it will never be possible, in any case for an AI system based on a digital computer – more precisely: mathematically equivalent to a Turing machine.
Nevertheless, many people nowadays regard it as virtually self-evident that computers will sooner or later become as intelligent as human beings – and, thereafter, become much more intelligent. The futurist Ray Kurzweil boosts his media ratings regularly with predictions to this effect.