Interpretability Illusion

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

• Interpretability as Natural Scienceframework0.806
  Proposed paradigm for evaluating interpretability work through empirical falsifiability rather than benchmarks or user studies
• Automated Interpretabilityframework0.800
  Method using large language models (Claude) to generate and test explanations of features at scale
• Circuit Interpretabilityconcept0.790
  Advantage of DiffLogic CA over NCA — learned rules are pure binary logic circuits that can be visualized and analyzed
• "For interpretability, I don't think we even have the right definitions."quote0.781
  Ian Goodfellow quote used to illustrate the pre-paradigmatic state of interpretability research
• Bottom-up interpretabilityconcept0.780
  An interpretability paradigm that explains computation in the model's own terms, rather than imposing top-down abstractions; VPD aims to realize this.
• Neural Network Interpretabilityconcept0.774
  The field aimed at understanding what neural networks have learned; characterized as pre-paradigmatic in this paper
• Interpretive Validationconcept0.771
  CIMC's methodology for evaluating whether a built system is conscious: combining multiple forms of evidence including predicted functional organization and developmental trajectories
• Curse of Dimensionality in Interpretabilityconcept0.759
  As models grow larger, the latent space volume grows exponentially, making enumeration impossible without decomposition