Anthropic published research revealing that its Claude language models have spontaneously developed an internal structure that mirrors one of the most influential theories of how human consciousness works. The researchers used a new mathematical technique to peer inside Claude's neural network and discovered what they call a "J-space", a small, privileged zone of internal activity where the model holds concepts it can report on, reason with, and direct at will, surrounded by a much larger ocean of automatic processes. The experiments were inspired by global workspace theory, developed to explain conscious access in the brain; a piece of information becomes consciously accessible when it gains entry to a small shared channel, the "workspace", which is broadcast to other brain systems.
When researchers prevented Claude from using its J-space, it still interacted normally, but lost its higher-order cognitive functions. However, Anthropic notes this tells us nothing about whether Claude is conscious in the way people are or whether it feels anything; their experiments do not show Claude can have experiences in the way humans do. Experts urge caution, stressing that it is premature to claim AI consciousness based on this finding.
The gap between what Claude does and what Claude feels remains exactly where it was: unmeasurable.