Artificial intelligencefromTechCrunch3 months agoA new, challenging AGI test stumps most AI models | TechCrunchThe new ARC-AGI-2 test challenges AI models with puzzle-like problems to measure their general intelligence more effectively than previous tests.