Eli Lilly CEO says he has 'at least 1 or 2 AIs running' during every meeting he's in (www.businessinsider.com)

Eli Lilly CEO David Ricks says he now runs “at least one or two AIs” during every meeting to keep up with medical literature and science updates, favoring Anthropic’s Claude and xAI’s Grok over OpenAI’s ChatGPT because he finds them terser and their references “check out more often.” He cautioned that hallucinations and bad citations remain a problem, so he still manually verifies outputs. The comment drew public notice (Elon Musk replied on X), and follows a trend of other CEOs—like Satya Nadella and Jensen Huang—noting routine AI use for summarization and tutoring. Ricks also spelled out why current models aren’t yet transformative for drug discovery: they lack a sufficiently comprehensive biological training corpus. He estimates humans understand only ~10–15% of biology today and believes models won’t be “good” until training data covers well above 50%. Achieving that, he argued, would require massive investment in automated experiments—24/7 robotics to generate large labeled datasets—a near-infrastructure-scale effort he thinks institutions like the NIH should lead. The remarks highlight two practical takeaways for AI/ML: executives are adopting AI for real-time scientific workflows, but biomedical applications face fundamental data gaps and persistent model failure modes (hallucination, unreliable citations) that demand coordinated data-generation and validation efforts.