Google has developed a new AI system to act as a 'co-scientist' and accelerate scientific discoveries by aiding human researchers.
The AI assistant can generate hypotheses, outline experiment plans, and synthesize existing literature in collaboration with scientists.
The system, named as an AI co-scientist, is built on Google's advanced AI models like Gemini 2.0.
It does not just summarize facts but aims to propose new hypotheses based on evidence, helping to generate original knowledge.
Google focuses on science applications for AI, following successes like AlphaFold, which solved the protein folding puzzle using AI.
The AI co-scientist comprises specialized AI programs that work together in steps resembling the scientific method.
Human scientists remain involved throughout the process, providing feedback and adjusting parameters as needed.
Google's AI system is designed to be collaborative, allowing scientists to insert their own ideas during the AI's cycle.
By outsourcing tasks like literature reviews and brainstorming to the AI co-scientist, researchers aim to speed up discovery and address grand challenges in science and medicine.
The AI system showed promising results in trials related to liver fibrosis and antibiotic resistance, indicating its potential to learn, refine solutions, and connect insights autonomously.