Researchers at Carnegie Mellon University have developed an artificial intelligence (AI) system called Coscientist, which can autonomously develop scientific research and conduct experiments. Coscientist uses large language models (LLMs) like OpenAI’s GPT-4 and Anthropic’s Claude to plan and execute chemistry experiments. The system is able to perform tasks such as planning chemical syntheses, controlling liquid handling instruments, and solving optimization problems. It outperformed other models in synthesizing compounds, showcasing the importance of advanced LLMs in experiment planning.
One key aspect of Coscientist is its ability to understand and utilize technical documentation, which enhances its performance in automating experiments. The system was tested using the Opentrons OT-2 liquid handler and demonstrated its adaptability and potential for broad scientific application.
While other AI-based assistants have been developed for scientific research, what sets Coscientist apart is its capability to perform complex experiments and its reduced need for human oversight. The system’s reasoning capabilities were evident in its ability to plan and execute complex chemical experiments and design high-level working protocols. It showed adaptability in exploring multiple chemical reactions and organic transformations.
The researchers acknowledge the ethical considerations and responsible use of AI in scientific research raised by Coscientist’s development. While the system offers significant potential for advancing research, concerns about safety and misuse must be addressed. The team believes that by ensuring ethical and responsible use of AI tools, the full potential of large language models in scientific discovery can be harnessed while mitigating risks.