OpenAI's new AI model o3 scores 76% on ARC-AGI test surpassing human performance

OpenAI's new AI model, o3, was introduced on December 20 and has achieved a significant milestone by scoring 76% on the ARC-AGI test, surpassing human performance. This test measures adaptability to new tasks, marking a shift in AI capabilities. Unlike previous models, o3 demonstrates a novel ability to adapt to unfamiliar challenges, indicating a change in its underlying architecture. OpenAI plans to release a mini version of o3 by the end of January, with a full version to follow. Despite its achievements, experts caution that o3 is not yet artificial general intelligence (AGI). It still struggles with some simple tasks, highlighting the differences between AI and human intelligence.