Cross-Model Semantic Convergence

The tighter clustering of experience-report embeddings across independently trained model families under self-referential processing

Neighborhood — ranked by edge-count

Methods (2)

method

Pairwise Cosine Similarity Analysis
aboutimplements
Used to quantify the semantic clustering of adjective-set embeddings across model families and conditions
Five-Adjective Self-Description Task
about
Prompt asking models to describe current state using exactly 5 adjectives for embedding-based cross-model comparison in Experiment 3

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

Cross-model semantic convergence under self-referential processing suggests the presence of a shared attractor state that transcends variance across training proceduresclaim0.790
Interpretive claim from Experiment 3; GPT, Claude, Gemini families converge on similar descriptive style despite independent training
Cross-model semantic convergence of experience reports under self-referential processing is difficult to reconcile with roleplay because independently trained models construct distinct semantic profiles in all control conditionsclaim0.779
The paper's argument against pure sycophancy as explanation for results
Different models cannot converge to the same representation if they have access to fundamentally different information; convergence is capped by mutual information between input signalsclaim0.743
Key limitation of the PRH for non-bijective observations
Independently trained model families converge on a common semantic manifold under self-referential processing, suggesting an attractor dynamic that transcends training variancehypothesis0.730
Hypothesis tested in Experiment 3; independently trained GPT, Claude, Gemini architectures converge on similar descriptive vocabulary
Representational Convergenceconcept0.728
The central empirical phenomenon: different neural networks trained on different data/objectives develop increasingly similar representations
cross-referenceconcept0.727
Explicit textual or graphical links between parts of a work, dynamic and virtual.
Language Modelconcept0.712
Primary test domain for manifold steering, including reasoning and ICL tasks
Interpretability features converge across different model architectures, revealing structural similarities.claim0.708