No Priors: AI, Machine Learning, Tech, & Startups · Nov 12, 2026 · 39m · 34m saved
With Max Jaderberg, President at Isomorphic Labs
Photo via isomorphiclabs.com
Drug approvals have stayed roughly flat for decades even as the cost of bringing one medicine to market rose from $100-300M to $1.5B, and timelines stretched from eight or nine years to fifteen. Most of that money is spent testing drugs in people, Max Jaderberg, president of Isomorphic Labs, told Elad Gil on No Priors, because almost nothing goes wrong until a drug actually reaches a patient.
"Where things really fail is in the clinic... we never could see the problems until we get to the point where we put these drugs in patients." — Max Jaderberg
Max helped build an ImageNet-winning computer-vision model in 2014, then joined DeepMind to "solve AGI," working on AlphaGo and AlphaStar before AlphaFold's early results convinced him and Demis Hassabis that deep learning could generalize to biology, which became Isomorphic Labs in 2021.
The full interview is covered here so you can skip it. 39 minutes of audio, 5 minutes of reading.
Here are the 6 insights that matter.
Key Takeaways
Drug approvals have stayed flat even as cost per drug rose to $1.5B and timelines stretched to 15 years
Animal testing is only about 30% predictive of human clinical toxicity, by Max's estimate
Isomorphic's goal is experimental accuracy in prediction, using the real world mainly to validate, not to learn
The company does not sell or license its drug design engine to anyone
A working pod runs roughly five AI researchers to one domain expert
Precisely engineered molecules could make multidrug cocktails safer, the way HIV is treated today
1. From Vision to Isomorphic
Max started a computer-vision PhD at Oxford just before AlexNet, and built an ImageNet-winning object-recognition network in 2014. When fundraising for a computer-vision startup proved hard, he approached Demis Hassabis for angel money and was invited to join DeepMind instead.
DeepMind recruited him to work on AGI, not to fund his startup
"This was probably 2018, '19... we were still trying to prove to the world that AI was even a thing." — Max Jaderberg
Max worked on generative models, deep reinforcement learning and game-playing systems like AlphaGo and AlphaStar before AlphaFold's results convinced him deep learning could generalize to biology.
"That was really the proof of concept that gave us conviction that now is the right time to go after... solving all disease." — Max Jaderberg
