The AI in Business Podcast Connecting the Dots Across Discovery - with Ben Ninio of Deloitte
8 snips
Oct 10, 2025 Benjamin Ninio, Principal in Strategy at Deloitte, shares his expertise in AI-driven scientific discovery. He discusses how language serves as a unifying framework across various R&D sectors like pharmaceuticals and agriculture. Ninio highlights the innovative role of large language models in reducing the need for heavy computation, facilitating the discovery of new insights. He also addresses the challenges of AI predictions, emphasizing the importance of validation and data quality while projecting an exciting future for AI in research.
AI Snips
Chapters
Transcript
Episode notes
Language As A Unifying R&D Framework
- Language serves as a unifying framework across R&D domains like pharma, agriculture, and chemistry.
- Using language models lets teams approximate complex natural systems without brute-force quantum computation.
Avoiding Brute-Force With Predictive Language
- Brute-force modeling of molecules quickly becomes computationally infeasible because biological systems are quantum and combinatorial.
- LLMs provide a 'hack' by predicting plausible next states in these languages, reducing compute needs.
Molecule Combinations Hit Compute Limits
- Modeling even two or three molecules explodes the search space into astronomic possibilities that outstrip compute.
- Benjamin likens this limit to hitting the practical end of brute‑force computational approaches.