New ‘AI scientists’ are improving – but reveal their fundamental limits
New ‘AI scientists’ are improving – but reveal their fundamental limits
Publish Date: 2026-05-20 02:26:00
Source Domain: theconversation.com
-
Specialised Connections in Science: Exciting scientific discoveries often require combining highly specialised knowledge with broad reasoning strategies, making connections between seemingly unrelated facts.
-
AI in Scientific Research: Artificial intelligence is increasingly being used to support scientific research by helping generate ideas, review existing work, and analyse data.
-
Language Model Use: New AI systems use large language models to allow scientists to interact naturally with vast bodies of scientific literature, enhancing the discovery process.
-
Limitations of Language-Based AI: While language-based AI can interact with textual scientific knowledge, it struggles with the precise and specific nature of scientific reasoning and the complexity of natural systems.
-
AI Systems in Development: Two new AI systems, Robin and Co-Scientist, aim to assist scientists more effectively in the discovery process by using multi-agent architectures capable of performing specific scientific tasks.
-
Impacts on Scientific Ecosystem: Initial concerns about how AI is impacting the scientific ecosystem are validated, showing mixed results in terms of paper quality and peer review standards.
-
AI Collaboration for Drug Repurposing: Robin and Co-Scientist have shown promise in drug repurposing tasks by identifying drug candidates, although they rely on human input for validation and testing.
-
Future of AI in Science: The next generation of AI models seeks to combine strengths from both data-driven and language-based approaches to ground scientific reasoning in structured knowledge, bridging the gap between words and the complexities of scientific phenomena.
