arXiv AI
Jun 12
Order Is Not Control
arXiv:2606. 12923v1 Announce Type: cross Abstract: AI alignment, interpretability, steering, and neural perturbation studies identify order-inducing objects.
Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability
arXiv:2608. 19338v1 Announce Type: cross Abstract: Mechanistic interpretability seeks quantities that models do not expose directly: represented states, component effects, interactions, and responses to interventions.
Related stories
arXiv AI
Jun 15
Order Is Not Control: Driven-Dissipative Response Laws Across Artificial and Biological Systems
arXiv:2606. 12923v2 Announce Type: replace-cross Abstract: AI alignment, interpretability, steering, and neural perturbation studies identify order-inducing objects.
arXiv Machine Learning
Jun 30
Internal-State Probes Read the Situation, Not the Action: Three Negative Results for Pre-Action Misalignment Monitoring
arXiv:2606. 30449v1 Announce Type: new Abstract: Probes on model internals could help monitor agentic systems if they identify harmful text or tool actions before those actions are generated.
arXiv Machine Learning
Jul 14
Causal Optimizer Interaction Calculus: Hidden Geometric Relaxation and Identifiable Interventions
arXiv:2607. 07206v2 Announce Type: replace Abstract: Optimizer experiments observe responses to algorithmic configurations without uniquely revealing hidden mechanisms.
arXiv AI
Jun 18
SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior
arXiv:2606. 18322v1 Announce Type: cross Abstract: Sparse Autoencoders (SAEs) decompose residual-stream activations into interpretable features.
arXiv AI
Jul 23
Reading and Steering Representations of Materials-Science Mechanisms in an Open-Weight Language Model
arXiv:2607. 20058v1 Announce Type: new Abstract: Large language models can answer scientific questions, yet a correct output does not reveal whether the model represents or uses the governing physics.
arXiv AI
Jun 2
Causal state binding predicts action control in language agents
arXiv:2605. 09692v3 Announce Type: replace Abstract: Autonomous language agents increasingly expose traces, memories, plans and constraints, but existing evaluations rarely test whether these state variables are bound to final actions.
arXiv AI
Aug 14
Decomposition of Evidence, Contradiction, and Fragility in Perturbation Responses
arXiv:2608. 12935v1 Announce Type: new Abstract: Perturbation methods explain model decisions by measuring prediction changes under altered inputs, but response magnitude tells us only how much a model reacts, not what that reaction means.
arXiv Machine Learning
Aug 12
Intrinsic Structure: Spectral Identifiability for Mechanistic Interpretability
arXiv:2608. 10172v1 Announce Type: new Abstract: Mechanistic interpretability explains models by identifying circuits inside them, but has no way to tell whether a circuit is a property of the model or an artifact of the method that found it.
arXiv AI
Aug 17
A Calibrated Test of Internal Action Maps: State Signals Without Global Affine Closure
arXiv:2608. 13626v1 Announce Type: new Abstract: A hidden state signal can be decodable or causally usable without supporting a reusable action map.
arXiv Machine Learning
Jul 30
What Can Latent World Models Know? Physical Parameter Identifiability in Multimodal Predictive Representations
arXiv:2607. 27017v1 Announce Type: new Abstract: A central premise of latent world models is that predicting the future forces a representation to internalize the physics of its environment.
arXiv Machine Learning
Aug 5
Sensitivity, Causality, and Repair Dissociate: A Layer-Wise Analysis of Perturbation Robustness and Its Scaling
arXiv:2608. 03842v1 Announce Type: cross Abstract: When a language model fails on surface-perturbed input (typos, OCR noise, homophones), "which layer is responsible" has three natural operationalizations: where representations diverge most (sensitivity), where restoring clean activations recovers the prediction (causality), and where a small adapter can repair the damage (compensatory capacity) - and we show these three layer maps dissociate.