AI
🔬 The Lab of the Future Should Feel Like a Data Center - Andy Beam & Rafa Gómez
Lila Sciences is building an automated, AI-guided robotics lab designed to act as an infinite token generator for scientific reasoning.
Key takeaways
- Lila Sciences has built a library of over 10 trillion experimentally validated scientific reasoning tokens.
- Science, not the internet, is the last untapped source of training data.
- A general scientific model beats domain-specific models sample for sample.
- Rafa's team rebuilt a gas sorption measurement to run roughly 2,500x faster.
Lila Sciences is building an automated, AI-guided robotics lab designed to act as an infinite token generator for scientific reasoning. Led by CTO Andy Beam and CSO Rafa Gómez-Bombarelli, the company has compiled a library of over 10 trillion experimentally validated scientific reasoning tokens. Lila Sciences aims to develop a scientific superintelligence by running biology, chemistry, drug discovery, and materials science experiments simultaneously in a single automated facility.
In their words
“you cannot make the ribosome go faster”
By the numbers
- 10T+
- Experimentally validated scientific reasoning tokens built by Lila
- 2,500x
- Speed increase for gas sorption measurement rebuilt by Rafa's team
- $2.1B
- Amount AbbVie paid for Capstan based on preclinical data
- 5%
- Mean FLOP utilization during RL training runs
Turn stories like this into views
Ravenclip finds the AI news, makes the video, and posts it before attention moves on.
#United States
#Lila Sciences
#Andy Beam
#Rafa Gómez-Bombarelli
#Escalante Bio
#Emily Whitehead
#Ken Stanley
Common questions
- What happened with Lila Sciences?
- Lila Sciences has built a library of over 10 trillion experimentally validated scientific reasoning tokens.
- Where can I read the original report?
- Read the full report at latent_space.
