LLM models are already capable of diagnosing scientific outputs, but, until now, had “no physical agency to actually perform" experiments. © 2024 TechCrunch. All rights reserved. For personal use only.