GPT-3 transforms chemical research (phys.org)
TLDR; Summary
- Artificial intelligence, particularly machine learning, is becoming crucial in chemical research
- Machine learning requires large amounts of data, which can be a limitation in chemical research
- Scientists at EPFL have used GPT-3, a large language model, to simplify chemical analysis
- GPT-3 is fine-tuned with a small dataset of questions and answers to provide accurate chemical insights
- The approach developed by EPFL researchers is simple, fast, and highly accurate, outperforming traditional machine learning models
- This method could revolutionize how chemical research is conducted, making it as easy as conducting a literature search
- Querying foundational models like GPT-3 could become a routine way to bootstrap projects in chemistry
- The study suggests that this approach will change the way chemistry is done