Foundries vs Navigators: Lowering the Cost of Science
In science, AI made thinking cheap but experiments remain the bottleneck, forcing biotech to adapt.
“Thinking got cheap and doing did not”— Dario Amodei
A Latent Space guest post argues that while LLMs have massively accelerated knowledge work and coding, scientific progress is still gated by slow physical experiments, so biotech is splitting into 'Foundries vs Navigators' to cope. It matters because it reframes the AI-for-science narrative around the experiment throughput bottleneck rather than pure model capability, alongside Anthropic's own science ambitions and new wet lab.