OpenAI’s new o3 artificial intelligence model has surpassed expectations by achieving a high score on the prestigious ARC Challenge, a test that measures an AI’s reasoning abilities. While this achievement is significant, experts emphasize that it’s just one step towards true artificial general intelligence (AGI). François Chollet, creator of the ARC Challenge, described o3 as a “surprising and important” milestone, highlighting its novel task adaptation capabilities. The o3 model builds upon large language models like ChatGPT, demonstrating a substantial increase in AI capabilities.
Source: https://www.newscientist.com/article/2462000-openais-o3-model-aced-a-test-of-ai-reasoning-but-its-still-not-agi