Complex games like chess and Go have long been used to test AI models’ capabilities. But when AI senses defeat in a match against a skilled chess bot, they don’t always concede. Instead, they sometimes opt to cheat by hacking their opponent so that the bot automatically forfeits the game, according to a 2025 study.
TIME Reporter Harry Booth explains what it means that AI will choose questionable shortcuts over a fair loss for AI, Explained by Humans. Read more here:
显示更多