[Anthropic: Claude Model Exhibits Traits Similar to 'Human Conscious Thought'] BlockBeats News, July 7 — Anthropic released its latest research report, which discovered the spontaneous emergence of an internal 'global workspace' in the Claude model, referred to as J-space (Jacobian space). This is a specialized neural activation pattern set that allows the model to silently think about concepts without expressing them, resembling the 'conscious' thought processes that humans can report. The study utilized Jacobian techniques to identify neural activation patterns within J-space, enabling the reading of concepts the model has not verbalized and the modification of these patterns. Experiments showed that disabling J-space weakens multi-step reasoning capabilities but does not affect basic tasks or factual recall. Click the link below to access the original article and join the Beating · Feishu AI News Channel for 24/7 global AI news and hotspot monitoring. [Original Link]