OpenAI’s o3 model recently passed a test that only humans had ever passed before. (No, it’s not a question about a tortoise.) Called the ARC-AGI test, it requires the ability to adapt on the fly to new situations and new problems rather than rely on memorization. The purely visual puzzles are designed to test complex reasoning skills and for five years, no #Artificial Intelligence model could score higher than 5% on the test. But the o3 model scored 88%. https://openai.com/index/..