(June 11, 2024, 1:34 PM EDT) -- HAMBURG, Germany — Even powerful artificial intelligences confidently provide the wrong answer to a basic reasoning problem easily solved by children and produce incorrect but “logic-like” explanations for incorrect responses, showing a “striking breakdown” of large language models (LLMs) suggesting that the models’ abilities are overblown, researchers say in an unreviewed paper published on the preprint website arxiv.org....