Breaking News
admin
Apr 22 2025
2During the April 2025 Bio-IT World Venture, Innovation, and Partnering conference, a panel comprising experts from leading technology and pharmaceutical companies convened to discuss the evolving role of artificial intelligence (AI) in drug development. The panel featured Rory Kelleher of NVIDIA, Bill Fitzgerald from Google Cloud, Bill Mayo of Bristol-Myers Squibb, Anthony Philippakis of GV (formerly Google Ventures), and Becky Stevenson of HSBC, with moderation by Jeremy Goldberg of Arsenal Capital Partners.
Advancements in Computing Power and AI Applications:
Rory Kelleher introduced NVIDIA's newly launched Blackwell computing platform, describing it as the most advanced AI accelerator available for training and deploying AI models at scale. He elaborated on three key "scaling laws" propelling AI innovation: pre-training for broad knowledge acquisition, post-training for domain-specific learning, and test-time compute for intensive reasoning. Kelleher emphasized the potential of these advancements in biological contexts, noting AI's capability to analyze scientific literature, formulate hypotheses, and design in silico experiments, thereby enhancing scientific productivity.
The Imperative of Understanding Causal Biology:
Bill Mayo underscored the necessity of comprehending causal human biology in pharmaceutical applications of AI. He cautioned that without accurate biological targets, even the most sophisticated models for protein folding, docking, or multi-parameter optimization would be ineffective. Mayo expressed concern that technology companies might inadvertently "solve biology," potentially relegating pharmaceutical firms to the role of contract research organizations. This possibility motivates Bristol-Myers Squibb to maintain a competitive edge in biological understanding while integrating AI tools.
Challenges in Data Generation and the Need for Partnerships:
The panelists concurred that generating high-quality biological data remains a significant obstacle. Kelleher advocated for public-private partnerships to develop the necessary datasets, suggesting that understanding the true causal relationships in biology might require data from a trillion cells. Bill Fitzgerald highlighted the importance of collaboration and tooling, noting that Google trains on publicly available datasets at a massive scale. He mentioned Google's release of several open-source healthcare AI models, including those for pathology and drug repurposing, aimed at accelerating development within the biotech community.
Investment Landscape Post-COVID:
Rebecca Stevenson provided insights into the current investment climate, describing a "post-COVID recovery period" where valuations for later-stage companies have declined, while early-stage companies are experiencing more favorable valuations. She pointed out the differing perspectives between tech investors, who often see limitless potential, and life science investors, who focus on de-risking around regulatory constraints such as FDA approvals.
The Human Element and Talent Shortage:
Anthony Philippakis identified the shortage of interdisciplinary talent as a major challenge. He emphasized the need for individuals who can bridge biology and computer science, making complex connections across disciplines. Philippakis noted that training such talent is currently the biggest rate-limiting step, as universities often lack the resources to develop models with the necessary parameters. He predicted that while pure mathematics might be the first field dramatically transformed by AI, drug development would be more challenging due to longer iteration cycles and the difficulty in establishing measurable benchmarks.
Looking Ahead: Predictions and Optimism:
Despite current uncertainties, the panelists expressed optimism about the future. Mayo anticipated that many technical challenges will be overcome within three years. Stevenson foresaw increased mergers and acquisitions as companies seek financing and advancement. Fitzgerald was optimistic about enhanced collaboration between personalized medicine, diagnostics, and pharmaceutical companies. Kelleher predicted that lab-in-the-loop and recursive, iterative learning would become industry standards, leading to faster target identification and reduced costs in early drug discovery.
As the biotech industry continues its AI-driven transformation, the panel emphasized that companies must decide whether to lead with science or technology. Those that successfully integrate both domains are poised to redefine drug discovery and development.
Source:https://www.bio-itworld.com/news/2025/04/22/ai-in-drug-development-promise-and-pitfalls
This is non-financial/medical advice and made using AI so could be wrong.
