Identifying Introspection From the Inside
This paper looks for mechanistic signs that language models can report faithfully on internal signals. It studies a controlled setting where self-report can be checked against known internal states.
From: Northeastern University; Eleos AI Research
The authors probe models to identify internal signatures associated with accurate self-report.
They report evidence that some forms of introspection can be detected mechanistically rather than only behaviorally.
The work does not establish that models are conscious, or that self-reports generalize to open-ended real-world settings.
Source: arXiv ↗. This is a summary; the source is the authority.
This modestly strengthens the case that some model self-reports track internal signals, but it does not show subjective experience.
How we weigh evidence like this: the evidence levels.
- Mechanistic Interpretability Should Prioritize Feature Consistency in Sparse AutoencodersThis paper argues that sparse autoencoders should be judged not only by reconstruction quality and sparsity but also by whether the same fea…
- Mechanistic Interpretability for Neural NetworksThis is a broad survey of mechanistic interpretability, covering the goal of reverse-engineering internal algorithms in neural networks. It …
- Constructing Interpretable Features from Compositional Neuron Representations in Language ModelsThis paper proposes an alternative to standard sparse autoencoders by decomposing MLP activations with semi-nonnegative matrix factorization…
- Interpretability Transfer from Language to Vision via Sparse Autoencoder Transfer AlignmentThis work transfers labeled sparse-autoencoder features from language to vision inside a multimodal model. It uses those features to make vi…