Interview, Fireside Chat
The Quest to ‘Solve All Diseases’ with AI: Isomorphic Labs’ Max Jaderberg
- The company aims to build a general AI-driven drug design engine applicable across all disease areas and modalities rather than targeting single indications, with the expectation that within five years, AI usage in drug design will become an industry inevitability as fundamental as mathematics.
- Achieving experimental-level accuracy for transformative drug design is predicted to require approximately a dozen "AlphaFold-like" breakthroughs, while generative models are expected to eventually explore a design space of 10^40 to 10^60 molecules.
- Future industry evolution includes potential "GPT-3" and "AlphaGo Move 37" moments where models exhibit superhuman creativity to generate uninterpretable yet physically correct designs, necessitating innovations in clinical trial structures and regulatory engagement to incorporate predictive toxicity and efficacy data.
- Significant opportunities exist to generate synthetic data via physics-based simulators and quantum chemistry theory, as well as to produce missing in vivo data using technologies like organ-on-a-chip to supplement or replace animal testing.
- Internal research programs will continue to develop models beyond structure to predict protein dynamics and cell function, while the industry is forecast to fully integrate AI to the point where the distinction between traditional pharmaceuticals and AI disappears.