Noah Smith: AI could usher in a golden age of science

Noah Smith argues that AI could reverse slowing scientific progress by expanding research capacity, speeding up theory and literature review, and finding patterns beyond human comprehension.

• More research in parallel: AI agents could tackle many questions without decades of human training.
• Faster theory: Smith predicts an order-of-magnitude reduction in theory turnaround times. This is a forecast, not an established general result.
• Large-scale literature synthesis: An OpenAI example cited in the article used thousands of agents to connect mouse-brain gene disruptions with published research. The team estimated comparable human analysis would take months to a year. The resulting leads still require experiments.
• New kinds of discovery: Smith speculates that AI could uncover useful regularities in complex systems, provided it receives enough physical-world data.

Implication: Faster hypothesis generation is not the same as faster validated discovery. Experiments, data collection and validation remain the key test.

Source: Noah Smith, Noahpinion, October 29, 2026. Cover: GPT-6-generated artwork from the source article.