Hallucination as a Latent Space Signal: Knowledge Boundary Detection and Cross-Model Routing via Latent Vectors
A conceptual note on interpreting LLM hallucination as a latent-space boundary signal rather than mere error. The work introduces the idea of hallucination as a “gradient of ignorance” and sketches an experimental protocol for boundary detection and cross-model routing via latent vectors.
Authors
- Pavel Scurin (ORCID: https://orcid.org/0009-0002-5299-4783)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-05-10
- DOI
- https://doi.org/10.5281/zenodo.20102211
- Primary Topic
- Adversarial Robustness in Machine Learning
- Type
- article
- Field-Weighted Citation Impact
- 0.00