Rather than pushing the epistemic edges, these AI-guided systems tend to excel at interpolating between existing data, not extrapolating to the radically novel and unknown. But at the very least, this silicon-based science could help reduce the epistemic load for carbon-based scientists and let them catch up to the frontier faster. And while cyborgs have yet to take over the world’s research labs, by fully automating replications to verify published findings, AI could help solve the reproducibility crisis.
