A set of puzzles that will problem even right this moment’s most subtle synthetic intelligence fashions, whereas being comparatively straightforward for folks, goals to encourage AI builders to create new methods. Any AI that solves the puzzles will internet its creators a share of a $1 million prize fund.
Companies like OpenAI already declare that their AI fashions, like GPT-4, exhibit “human-level performance” on real-world exams, similar to college admission exams or the bar examination for legal professionals. But this isn’t as a result of the fashions are reasoning intelligently like humans, says Mike Knoop at…