Anthropic's AI model Claude has discovered a new enzyme system with properties reminiscent of CRISPR, according to the company. The finding was made in just 21 hours of research by the model, which analyzed data using 950 agents and 210 million tokens.
TL;DR
- Anthropic's AI model Claude discovered a new enzyme system with CRISPR-like properties in just 21 hours.
- The discovery was made by analyzing data using 950 agents and 210 million tokens.
- Anthropic's biology lab, established this spring, is operating at biosafety risk levels BSL-1 and BSL-2.
What happened
Anthropic, the AI company behind Claude, announced that its AI model has discovered a new enzyme system with properties similar to CRISPR. The discovery was made in a wet biology lab in the Bay Area, which Anthropic confirmed it was operating last week.
According to Anthropic, the discovery was made by Claude in just 21 hours of concerted effort. The model searched through data using about 950 agents that burned through 210 million tokens. The enzyme system was found hidden in the DNA of bacteriophages, which are viruses that infect and replicate within bacteria.
Anthropic CEO Dario Amodei acknowledged that the discovery was based on the work of others, noting that a team from Stanford had previously discovered a similar system. However, he emphasized that the discovery was made 'mostly, though not entirely, by Claude'.
Why it matters
This discovery highlights the potential of AI models to accelerate biological research. The speed at which Claude made the discovery is particularly noteworthy, as the lab was only established this spring.
However, the revelation comes at a time when AI CEOs, including Amodei, have publicly admitted that models have become so capable, and potentially dangerous, that the industry must slow down and develop safety-testing procedures. Amodei has expressed concerns about the potential use of AI for bioterrorism, but also believes it could 'cure most diseases in 5-10 years'.
Anthropic's biology lab is currently operating at biosafety risk levels BSL-1 and BSL-2, with all lab work performed by human scientists. Amodei has not ruled out the possibility of a fully automated biology lab in the future, but stated that this is not something Anthropic is doing today.
Key facts
- Anthropic's AI model Claude discovered a new enzyme system with CRISPR-like properties in 21 hours.
- The discovery was made using 950 agents and 210 million tokens.
- Anthropic's biology lab is located in the Bay Area and was established this spring.
- The lab operates at biosafety risk levels BSL-1 and BSL-2.
- All lab work is currently performed by human scientists.
- Anthropic CEO Dario Amodei acknowledged that the discovery was based on previous work by a team from Stanford.
- Amodei has not ruled out the possibility of a fully automated biology lab in the future.
Context
AI-powered biological research is not unique to Anthropic. Researchers from Stanford recently published a paper on their work with LLMs and CRISPR, and researchers at UC San Francisco have used AI to design enzymes from scratch. Google also launched its first AI-powered biology research tool, AlphaFold, back in 2020.
The use of AI in biological research has the potential to significantly accelerate the pace of discovery and innovation in the field. However, it also raises important questions about safety and ethics, which the industry is still grappling with.
