Artificial-intelligence agents are moving from drafting text to drafting science. Two Nature studies showcase multi-agent systems that read literature, propose biomedical hypotheses, plan experiments and analyze results—hinting at an end-to-end discovery pipeline with AI at each step. Advocates say the approach could compress timelines for drug discovery and other lab work; skeptics note that validation, data quality and oversight remain decisive. The commentary flags financial ties between the author and multiple biotech firms, underscoring the sector’s commercial stakes as laboratories test AI’s promise against practical constraints.
Related articles:
Highly accurate protein structure prediction with AlphaFold
A mobile robotic chemist
ChemCrow: Augmenting large-language models with chemistry tools
AI systems devise hypotheses and ways to test them (Nature News & Views context)
