Transformer-based LLMs do not clearly satisfy the GWT indicator properties.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Sequences of contemporary Transformer-based LLM representations lack statistically significant indicators of observed 'consciousness' phenomena under the three stringent criteria.claim0.777
  Primary conclusion of the study based on temporal permutation analysis failing all three criteria.
• The problem isn't that it is a transformer. The problem is that it is an auto-regressive LLM. Auto-regressive LLMs that compute each token with a fixed number of computational steps can't reason, regardless of the details of the architecture.quote0.764
  LeCun's post on X supporting the view that fixed-step probabilistic prediction precludes consciousness in LLMs.
• Current LLMs cannot faithfully represent transformative experiences with epistemically opaque outcomes.claim0.751
• Automated interpretability using LLMs can usefully score feature specificity.claim0.742
  Claude 3 Opus ratings aligned with human judgment of feature descriptions.
• All cases satisfying Criteria 1 and 2 (two out of three) originate from deeper transformer layers and/or the 2/3 layer of LLMs.finding0.739
  Consistent with literature that deeper layers encode semantic information and align with human brain activity.
• So at any point in the network, the transformer not only receives information from its past... but also has causal influence over its future processing. So, saying that LLMs cannot introspect... is incorrect.quote0.737
  Core summary of Janus' position on autoregressive recurrence enabling introspection.
• Whether conclusions about latent reflection directions generalize to larger LLMs, different architectures, or broader datasets remains to be verified.question0.729
  Key limitation and open question about experimental scope.
• Transformer architecture permits introspection; saying LLMs cannot introspect on past internal states is wrong.claim0.725
  Janus's central claim that the architecture enables introspection, though usage in practice is a separate question.