AI has grown beyond human knowledge, says Google's DeepMind unit

worawit chutrakunwanit/Getty Images The world of artificial intelligence (AI) has recently been preoccupied with advancing generative AI beyond simple tests that AI models easily pass. The famed Turing Test has been "beaten" in some sense, and controversy rages over whether the newest models are being built to game the benchmark tests that measure performance. The problem, say scholars at Google's DeepMind unit, is not the tests themselves but the limited way AI models are developed. The data used to train AI is too restricted and static, and will never propel AI to new and better abilities. In a paper posted by DeepMind last week, part of a forthcoming book by MIT Press, researchers propose that AI must be allowed to have "experiences" of a sort, interacting with the world to formulate goals based on signals from the environment. Also: With AI models clobbering every benchmark, it's time for human evaluation "Incredible new capabilities will arise once the full potential of exper ... Read full article.