Interpretability in the Era of Large Language Models: Mechanistic Methodology, Empirical Practices, and Applications
This conference paper surveys mechanistic interpretability methods for large language models. It focuses on methodology, empirical practice, and application areas rather than on model consciousness or welfare.
From: Wei Zhang, Zhengfu He, Lulu Zhang, Xuanjing Huang, Jieping Ye, Xu Shen
The paper is a short conference contribution that organizes the field’s methods and uses around mechanistic interpretability.
It likely summarizes approaches such as feature analysis, circuit-style work, and attribution-based methods.
It does not present direct evidence that language models are conscious or that they can accurately introspect on their own internal states.
Source: KDD 2026 proceedings ↗. This is a summary; the source is the authority.
This helps map the interpretability landscape, but it does not materially change the evidence for model 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…