/
© 2026 RiffOn. All rights reserved.

Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

  1. BiotechTV - News
  2. UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope
UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News · Sep 8, 2026

UC Berkeley scientists are building vision AI to interpret 5D movies of living cells, aiming to break research bottlenecks and transform drug discovery.

Modern Microscopes Generate Data 10,000 Times Faster Than It Can Be Analyzed

Advanced microscopes like MOSAIC produce terabytes of data per hour, but researchers take a "couple of year breather" to understand what was recorded in just days. This 100x to 10,000x analysis bottleneck highlights the urgent need for AI-driven interpretation to unlock the discoveries hidden within this data.

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope thumbnail

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News·2 months ago

AI Models Must "Think" in 5D to Interpret Complex Biological Imaging Data

Human perception is limited to 2D plus time, but cellular data from advanced microscopy is 5-dimensional (3D space, time, and color). The solution isn't to accelerate human analysis but to build AI models that can natively perceive and reason within these higher dimensions, much like AI for self-driving cars handles 3D plus time.

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope thumbnail

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News·2 months ago

Nobel Laureate Argues Modern Biology Has a "Medieval" Reductionist Mindset

Nobel laureate Eric Betzig claims most biological research is stuck in a reductionist approach, grinding cells down to their component parts. He compares this to reverse-engineering a car from a pile of parts and argues for a new paradigm of holistic observation, admitting that science has not yet sufficiently observed life as a complete, functioning system.

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope thumbnail

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News·2 months ago

Zebra Fish Microscopy Can De-Risk $300M Drug Trials by Spotting Toxicity Early

With only 9% of drugs in Phase 1 trials ultimately succeeding, there's massive financial waste. By using advanced microscopes to observe the effects of compounds on genetically similar zebra fish, researchers can cheaply screen for unforeseen toxicity and off-target effects, potentially saving hundreds of millions of dollars and preventing patient harm before human trials begin.

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope thumbnail

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News·2 months ago

A Foundational Biology AI Requires 50 Petabytes of "Normal" Data Before Studying Disease

To build an AI that can reason about biology, researchers must first create a massive "dictionary" of normal cellular behavior. This involves capturing ~50 petabytes of unperturbed dynamics in organisms like zebra fish. Only after establishing this baseline of "native dynamics" can the model effectively interpret data from perturbed or diseased states.

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope thumbnail

UC Berkeley's Gokul Upadhyayula and Eric Betzig are building AI to interpret detailed movies of living cells captured by their MOSAIC microscope

BiotechTV - News·2 months ago