OpenAI's latest AI model, o3, has achieved a score of 87.5% on the challenging Autonomous Research Collaborative Artificial General Intelligence (ARC-AGI) benchmark, coming close to the minimum threshold for human-level performance. The model represents a breakthrough in complex reasoning, using a "program synthesis" approach to tackle new problems. However, experts have debated whether o3 truly represents AGI, with some arguing that it falls short and others suggesting that it has breached the benchmark. While OpenAI CEO Sam Altman does not take a definitive stance on the matter, he sees o3 as the beginning of the next phase of AI.



Other News from Today