> AI-driven labs can iterate 'good enough' hypotheses way faster than human R&D systems.

Is there evidence of that?