OpenAI has introduced GDPval, an evaluation metric designed to measure the performance of AI models on real-world tasks with economic value. The metric assesses model performance across 44 different occupations, aiming to provide a more comprehensive understanding of a model's capabilities. This is part of OpenAI's efforts to develop more practical and useful AI systems. --- Why it matters: This matters because it allows researchers to evaluate the actual usefulness of their models in real-world scenarios, rather than just on narrow benchmarks or datasets. Source: https://openai.com/index/gdpval