Hinton Urges Designing for Control and Coexistence, Not AI Labels

According to AI TIMES, Professor Geoffrey Hinton sharply criticized the political use of the term “superintelligence” in a Guardian interview introduced on October 14, saying the more urgent issue is not what to call AI, but how to control and coexist with a form of intelligence different from humans. He said he has long assessed the chance that AI could pose a serious threat to humanity at 10% to 20%, and warned that society does not have much time to adapt. In the article, Hinton views AI not as a simple tool but as an entity that learns in ways different from humans, raising a fundamental question about whether humans can continue to maintain control. He also argued that the possibility that large language models may understand things or even have consciousness should not be dismissed too quickly, expressing concern that the idea of humans as the center of intelligence itself needs to be reconsidered. He drew a line against approaches that rely on corporate self-regulation, saying that public oversight, safeguards, and accountability structures must be discussed alongside technological competition. In the end, the article argues that beyond competition over AI performance, the bigger issue is what values and relationships will be used to bind this technology to society.