Anthropic Probes Claude's Mysterious Mind
Source: newyorker.com
- $350 billion, runs quirky experiments like Claude managing a vending machine called Project Vend.
Anthropic, a leading AI firm in San Francisco, is at the forefront of "interpretability" research to understand how its popular chatbot Claude actually works and what it "is." Researchers treat Claude like a psychological subject, mapping its neurons and testing its behaviors in real-world simulations. The core finding is that both AI and human minds are black boxes we don't fully grasp, prompting a new scientific push to map AI's terrain. This matters because talking machines blur the line betw