Algorithm Antics: AI Models Take on Physics Problems
In the ever-evolving landscape of AI, it’s easy to get swept up in the hype surrounding the latest and greatest models. Recently, New Zealand Physics teacher Benny Pan shared an intriguing LinkedIn post where he put GPT4-o1, Mistral, and GPT-4o mini through their paces on a complex NCEA Level 3 physics problem involving the rotational inertia of a satellite. Inspired by Benny’s experiment, I decided to conduct my own test, this time focusing on a complex 2D collision calculation. Armed with Microsoft Co-pilot, Google Gemini, GPT-o1, and GPT-4o mini, I set out to see how these models would handle the challenge. Just like Benny, I used quality data and Chain-of-Thought (COT) prompt engineering techniques to ensure a fair and rigorous evaluation. So, fellow educators, buckle up and join me on this enlightening journey through the world of AI in education. Let’s explore how these powerful tools can enhance our teaching and inspire our students to reach new heights in their understanding