When Anthropic's interpretability team asked whether there might be concepts inside Claude that the model was actively entertaining without writing down, they built a mathematical tool to find out.

Anthropic says Claude developed a hidden thinking space called J-space on its own during training. J-space resembles the brain's global workspace, broadcasting information across the whole model.

This postdoctoral researcher develops innovative artificial intelligence and statistical methods to uncover hidden, reliable patterns within complex global datasets.

Customer stories Events & webinars Ebooks & reports Business insights GitHub Skills ...

A metadata commons to store research software metadata - arash77/research-software-ecosystem-content ...