Advanced AI model that achieved 96.7% accuracy on competitive mathematics benchmarks and 87.7% on PhD-level science questions, raising concerns about white-collar job displacement.
"The newest Chatgpt model O3 ( expensive to run, costs about 3000 dollars per query but the cost will be go down quickly) achieved an accuracy of 96.7% on the AIME 2024 benchmark, 83.3% for competitive mathematics. Similarly, in PhD-level science questions from the GPQA Diamond dataset, O3 reached an"