The Panel
Noubar Afeyan, co-founder and chairman of Moderna and founder and CEO of Flagship Pioneering, Lila Ibrahim, chief AI readiness officer at Google DeepMind, and Kiran Mazumdar-Shaw, founder and executive chairperson of Biocon, spoke together on stage in New Delhi. Bloomberg’s Francine Lacqua moderated the exchange on 13 October 2026. The session focused on the ways artificial intelligence is altering how researchers identify targets, design experiments, and move discoveries toward application.
Forum Context
The Bloomberg New Economy Forum brings together executives and scientists to examine technology’s effect on global industries. This year’s edition placed particular weight on life sciences and computing. The three participants represent distinct parts of the research pipeline: vaccine and therapeutic development at Moderna, large-scale machine-learning systems at DeepMind, and biologics manufacturing at Biocon. Their joint appearance signals that AI tools have moved from pilot projects into core scientific operations across these organizations.
The forum itself took place against a backdrop of rapid model releases and regulatory scrutiny over data use in health applications. New Delhi as host city underscored interest from emerging markets in how these technologies might affect local manufacturing and clinical research capacity. No new datasets or model releases were announced during the discussion.
What the Speakers Covered
Afeyan described how models now help Flagship and Moderna test hypotheses about mRNA constructs before any laboratory work begins. Ibrahim outlined DeepMind’s efforts to make AI systems reliable enough for regulated research environments, including reproducibility checks and uncertainty estimates. Mazumdar-Shaw discussed Biocon’s use of predictive models to shorten process development cycles in biologic drug production. The panel did not release new technical benchmarks or product timelines; instead, the speakers compared the practical constraints each organization still faces when deploying these tools at scale.
The conversation stayed at the level of operational experience rather than technical architecture. Speakers noted the need for traceable decision paths when models influence which compounds advance to animal or human testing. They also referenced the difficulty of aligning model outputs with existing quality-control standards in manufacturing.
Reactions and Limits
No other panelists or audience members offered direct rebuttals during the session. The absence of competing viewpoints left open questions about smaller organizations that lack comparable internal data or compute resources. The speakers did not address competitive dynamics between their companies or potential overlaps in their AI approaches.
Why It Matters
The conversation shows that AI adoption in science is no longer limited to early-stage screening. Companies that control both data generation and physical manufacturing are now integrating models into later stages of development and production. This integration changes the required skill set for teams working on biologics or mRNA programs: familiarity with model validation becomes as relevant as traditional assay design.
For engineers and researchers outside these firms, the takeaway is narrower. Progress still depends on access to high-quality experimental data and on regulatory comfort with model-driven decisions. Organizations that cannot generate or license comparable data will continue to lag. The panel offered no evidence that AI has shortened overall drug-development timelines yet; it only confirmed that the largest players treat the technology as a permanent part of their workflow.
Regulatory agencies have so far issued guidance focused on transparency rather than outright approval of model-only submissions. Until those rules stabilize, the practical advantage remains with companies that already run their own wet labs and can generate fresh data to retrain or audit models. Smaller research groups or contract manufacturers may find themselves dependent on the same few data-rich organizations for both tools and validation.
The panel also highlighted a quiet shift in hiring. Roles once described as “data scientist supporting R&D” now sit closer to core decision-making on which programs advance. This reordering of influence inside companies is likely to spread to academic collaborations and contract research organizations that serve them. The discussion did not resolve whether this concentration of capability will accelerate overall scientific output or simply raise the bar for participation.
---
Sources:
[
{
"publisher": "Bloomberg Technology",
"title": "Moderna, Google DeepMind, Biocon on AI in Science",
"url": "https://www.bloomberg.com/news/videos/2026-10-13/moderna-google-deepmind-biocon-on-ai-in-science-video",
"published_at": "2026-10-13T10:31:41.000Z",
"summary": "Noubar Afeyan, Co-Founder & Chairman, Moderna and Founder & CEO, Flagship Pioneering; Lila Ibrahim, Chief AI Readiness Officer, Google DeepMind; and Kiran Mazumdar-Shaw, Founder & Executive Chairperson, Biocon discuss how AI is reshaping scientific discovery with Bloomberg's Francine Lacqua at the 2026 Bloomberg New Economy Forum in New Delhi. (Source: Bloomberg)"
}
]
No comments yet