# When AI Designs a Drug, Who Gets the Credit?
The question of attribution in AI-driven drug discovery has moved from theoretical to urgent. Insilico Medicine's recent claims about its generative AI platform "discovering" a novel molecule for pulmonary fibrosis highlight a growing tension between marketing language and scientific reality.
Insilico leads a cohort of biotech companies weaponizing AI to accelerate drug design. The appeal is obvious. Traditional drug discovery takes years and costs billions. AI models can screen millions of compounds in hours, identify promising candidates, and predict how they behave in biological systems. The speed advantage is real. The attribution problem is thornier.
When Insilico declared its AI "discovered" a drug candidate, the company tapped into powerful framing. Discovery suggests novelty and insight. It credits the tool with intellectual achievement. But drug discovery involves multiple layers of work. Engineers built the AI system. Scientists curated training data. Chemists designed the prompts and evaluated outputs. Researchers at academic institutions or contract labs will likely conduct the actual wet-lab validation. Clinicians will run trials. The molecule's efficacy and safety remain unproven.
This matters because credit shapes funding, careers, and how society understands technological progress. If AI gets the headline, humans become invisible. If humans claim sole credit for work their algorithms enabled, that distorts the picture too.
The stakes ripple through the field. Regulatory bodies like the FDA care about reproducibility and accountability. If a drug candidate comes from an AI system, who takes responsibility when something goes wrong? Patent offices face unprecedented questions about inventorship. Can an AI system be named as an inventor on a patent? Different jurisdictions answer differently. Some countries allow it. Others don't.
The talent market also shifts with attribution clarity. Researchers choosing between academia and biotech weigh their ability to get published and cited. If AI systems absorb the discoverer credit, academic incentives warp. Young scientists may lose motivation to enter drug discovery if their contributions become invisible.
Beyond credit, attribution raises transparency questions. Insilico's generative AI operates as a black box to some degree. When the system proposes a molecule, explaining why it selected that specific compound becomes difficult. The AI might identify patterns humans miss, or it might exploit statistical quirks in training data. Scientists need to understand the reasoning behind AI recommendations to validate results independently.
Some companies frame this differently. They describe AI as a tool that augments human researchers rather than replaces them. This language acknowledges human expertise while claiming efficiency gains. It's more accurate but less headline-grabbing than "AI discovered."
The biotech industry hasn't settled on standard attribution language. Insilico's approach reflects aggressive positioning in a competitive field. As more AI-discovered drugs enter clinical trials, journals and regulatory bodies will likely establish clearer conventions. The first AI-discovered drug to reach FDA approval will crystallize these questions.
The principle emerging from this debate is simple. Attribution should reflect actual work and decision-making. AI systems are powerful tools that accelerate specific steps in drug design. They don't think, hypothesize, or take creative risks the way human scientists do. Crediting AI with discovery sells a more exciting story than crediting the researchers who built, deployed, and interpreted the AI. But accuracy matters more than narrative appeal when lives depend on drug safety.