In reported closed-loop systems, the machine typically owns the search: which candidate to try next, how to adjust the recipe, and when to stop a line of inquiry. Humans still own the framing decisions: which problem is worth solving, what constraints the result must satisfy, which measurements count as success, and how to interpret a surprising outcome. The split is not a limitation of the software alone; it reflects which decisions require judgment about purpose and consequence.
Learn MotionCan AI Run a Drug Discovery Lab on Its Own?
Real Examples of AI-Run Lab Systems
What the Machine Chose, What the Human Chose
2 / 3
So the loop runs, but who decided what inside it? Look at the choices the machine actually made. It picked which candidate to try next, it adjusted the recipe when a result came back weak, and it decided when a line of inquiry was no longer worth pursuing.
0:00 / 0:00