Stanford Researchers Debate Whether AI Can Guide Humans Through Wet Lab Experiments

On September 25, Stanford computational biologist and AI Index co-director Anshul Kundaje debated with researcher suragnair on X over the idea of "AI guiding humans through biological wet experiments." The core disagreement: whether the tacit skills involved in biological experiments can be taught by AI.

Confirmed

• suragnair's original vision: AI would not perform experimental steps itself, but learn how to optimally guide human operators, suggesting corrections after humans complete all or part of an operation, thereby progressively accumulating wet-lab capability.
• Kundaje challenged this: many biological experiments require dexterous hand movements and hard-to-articulate "feel." Humans learn by first watching experts and then trial and error—visual AI still struggles with the first step—and human novices without explicit motor control may not be able to execute operations from AI instructions alone.
• suragnair responded that an AI trained to observe everything would be more "passive," but in his framing, AI plays a guiding and corrective role rather than replacing human execution.

Why it matters

• The debate touches on the fundamental difference between embodied skill acquisition and transfer via language/visual instructions: whether tacit knowledge (feel) can be made explicit and effectively communicated is a key bottleneck for AI entering experimental automation.
• The two sides are not entirely opposed—the debate actually converges on "AI's role definition": if AI is positioned as a coach rather than an operator, many objections can be sidestepped, laying a groundwork for discussing the "AI coach + human hands" path to automating biological experiments.

2026-09-25 ~ 2026-09-25 · 5 related posts

• [source] Suragnair responds: AI's role in wet labs is to instruct humans, not execute — suragnair · 2026-09-25
• [source] Bioinformatician Anshul Kundaje doubts AI can talk novices through wet-lab experiments — anshulkundaje · 2026-09-25
• Researchers debate whether observation-trained AI can guide hands-on biology experiments — suragnair · 2026-09-25

2 near-duplicate retellings: anshulkundaje · suragnair