Capability denialism describes the claim that a given approach in Artificial Intelligence can never achieve human performance, often despite some evidence to the contrary. Capability denialism can take many forms, but it usually does not attempt any formal proof.