AI Chemistry

Chemistry involves complicated processes which make it challenging to automate. But cutting-edge artificial intelligence systems are being designed to be capable of even the most complex chemical formulas. We’ll explore how one AI system takes inspiration from Crows to do so in the U.S. National Science Foundation’s “Discovery Files”.
With vast potential for innovation, artificial intelligence systems are revolutionizing the world around us. An international partnership including NSF-supported researchers at the University of Rochester have developed ChemCrow, an AI system that integrates 18 expertly designed tools, enabling it to navigate and perform chemical research with unprecedented efficiency.
ChemCrow is based on a large language model, augmented with a suite of specialized software, including safety assessment guardrails and enhanced with tool integrations to autonomously plan and execute chemical synthesis tasks.
When ChemCrow receives a prompt from a user, it plans how to solve the task, selects the relevant tools, and refines its strategy along each step. This methodical approach ensures that theory is applied while being grounded in practical application for real-world laboratory environments.
ChemCrow lowers the barrier to entry for complex chemical knowledge and processes, while augmenting the advanced abilities of veteran chemists. This combination could accelerate research and development in pharmaceuticals, materials science, and beyond, while making the process safer and more efficient.
To hear more science and engineering news, including the researchers making it, subscribe to "NSF's Discovery Files" podcast.