How will we know if AI is smart enough to do science?

Key Points

• The research aims to determine how effectively AI can utilize knowledge to make scientific discoveries.
• Developed tests to assess large language models' performance
• Evaluated models on their ability to make findings
• Compared results across different models
• Models demonstrated varying levels of effectiveness in making discoveries
• Some models identified novel insights, while others struggled
• Establishing benchmarks for AI's capabilities in science