Evaluating ChatGPT on 1,000 Math Word Problems

Explore an in-depth study evaluating ChatGPT's performance on 1,000 math word problems in this 16-minute video presentation. Delve into multiple experiments designed to understand the capabilities and limitations of OpenAI's commercial large language model. Learn about the findings from the paper "An Independent Evaluation of ChatGPT on Mathematical Word Problems (MWP)" presented at the AAAI Spring Symposium. Gain insights into the intersection of symbolic methods and deep learning, and discover how this research contributes to the broader field of artificial intelligence and machine learning. Access the preprint of the paper for further details and explore the Neuro Symbolic Channel for more content on cutting-edge AI topics, including tutorials, courses, and research results aimed at advancing artificial general intelligence (AGI).