latent-knowledge

Tag

Cards List
#latent-knowledge

The Impossibility of Eliciting Latent Knowledge

arXiv cs.AI · 2026-06-11 Cached

This paper formally defines the problem of eliciting latent knowledge (ELK) from AI systems using Causal Influence Diagrams, and proves an impossibility theorem: no feedback-based training strategy that depends only on agent behavior can guarantee an honest agent, even with perfect training feedback.

0 favorites 0 likes
#latent-knowledge

Forgetting is Not Erasure: Recovering Latent Knowledge via Transport Keys

arXiv cs.LG · 2026-06-03 Cached

This paper argues that catastrophic forgetting in neural networks is not erasure but an interface alignment problem. It introduces 'transport keys' to recover latent task-specific features from sequentially trained models, demonstrating significant performance recovery on split CIFAR-100.

0 favorites 0 likes
#latent-knowledge

MechELK: A Mechanistic Interpretability Framework for Eliciting Latent Knowledge in Large Language Models

arXiv cs.CL · 2026-05-29 Cached

MechELK is a three-stage framework combining mechanistic interpretability tools (SAE, activation patching, causal probing) with representation engineering to elicit latent knowledge from LLMs, achieving 84.7% accuracy and outperforming existing methods like CCS and linear probing.

0 favorites 0 likes
← Back to home

Submit Feedback