William Poundstone reflects on the question of whether machines can think and the relevance of that question. He likens it to asking whether submarines can swim, emphasizing that the focus on how closely machine intelligence can duplicate human intelligence might not be the real point.
Poundstone acknowledges that machine intelligence can take various directions and that computers excel at tasks humans struggle with. He suggests that it’s a failure of imagination to solely focus on human-like directions for AI development.
While he acknowledges that there might not be a practical need for future machines to have emotions, inner dialogue, or the desire for legal and civil rights, he acknowledges that some people will desire anthropomorphic machine intelligence for various reasons, including aesthetics and publicity.
He raises the idea that as technology advances and human-like machine intelligence becomes more accessible, there may be interest in creating machines with their own will and interests, potentially diverging from human values. Poundstone also suggests that the notion of AI turning on its creators, like in the Frankenstein story, is worth taking seriously in the context of AI development.