Mechanistic interpretability
Publication dates and source age
Sources counted: 4
Newest dated source: 2025-07-31
At least one source was published within the 18-month window.
Assessed at this edition's evidence cutoff: 2026-09-15. 18-month boundary: 2025-03-15.
Publication age does not tell us whether a claim is still valid. Reading an old source again does not make its publication date newer. An update date does not establish that the passage we used was updated.
Open in the glossary Reading notes · Structured record
In plain language
Research into how a model's learned internal computations produce its behavior. [1]
Reference this explanation or suggest a correction
Link to this explanation · Suggest a correction · How corrections work
Limits & distinctions
Model-generated chain of thought is an incomplete record, not a complete account of internal computation. It may still help monitoring alongside evidence about actions and other safeguards. [4]
Reference this explanation or suggest a correction
Link to this explanation · Suggest a correction · How corrections work
A fuller explanation
Mechanistic interpretability tries to identify understandable features and the computations connecting them. Circuit-tracing work studies selected mechanisms and tests proposed explanations through interventions. [1] [2]
Reference this explanation or suggest a correction
Link to this explanation · Suggest a correction · How corrections work
How it relates to the map
An explanation can inform evaluation without deciding a development policy or certifying every deployment of the model. [3] [1]
Reference this explanation or suggest a correction
Link to this explanation · Suggest a correction · How corrections work
Share this page
https://theaiatlas.org/ideas/interpretability/
Download a share image · Vector image
Image previews are summaries. Keep the page link so readers can check the evidence.
Sources and what we read
1. Circuit Tracing: Revealing Computational Graphs in Language Models
Publication dates and source age
Sources counted: 1
Newest dated source: 2025-03-27
At least one source was published within the 18-month window.
Assessed at this edition's evidence cutoff: 2026-09-15. 18-month boundary: 2025-03-15.
Publication age does not tell us whether a claim is still valid. Reading an old source again does not make its publication date newer. An update date does not establish that the passage we used was updated.
Read introduction, method overview and limitations including reconstruction errors, graph complexity, global circuits and mechanistic faithfulness. The authors' replacement-model analyses reveal selected mechanisms; they do not provide a complete explanation of all behavior. Later attention-tracing work is cited alongside this paper to avoid treating its missing-attention limitation as a permanent field-wide result.
2. Tracing Attention Computation Through Feature Interactions
Publication dates and source age
Sources counted: 1
Newest dated source: 2025-07-31
At least one source was published within the 18-month window.
Assessed at this edition's evidence cutoff: 2026-09-15. 18-month boundary: 2025-03-15.
Publication age does not tell us whether a claim is still valid. Reading an old source again does not make its publication date newer. An update date does not establish that the passage we used was updated.
Direct browser-tool retrieval failed; fetched the original publisher HTML successfully and read the introduction, case-study summaries, QK-attribution method, inhibitory-effect limitation and graph-construction tradeoffs. Extends earlier attribution graphs to attention; results are selected studies with open questions, not a complete model explanation.
3. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
Source is over 18 months old.
Publication dates and source age
Sources counted: 1
Newest dated source: 2024-07
Assessed at this edition's evidence cutoff: 2026-09-15. 18-month boundary: 2025-03-15.
Publication age does not tell us whether a claim is still valid. Reading an old source again does not make its publication date newer. An update date does not establish that the passage we used was updated.
Read introduction, section 2.2 on confabulation, and selected MEASURE actions 2.3, 2.5, 2.6, 2.7 and 2.9 concerning evaluation evidence, generalization, citations, generated-code review and safeguards. A voluntary risk-management profile; no claim that all 64 pages or every referenced study was reviewed.
4. Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
Publication dates and source age
Sources counted: 1
Newest dated source: 2025-07-15
At least one source was published within the 18-month window.
Assessed at this edition's evidence cutoff: 2026-09-15. 18-month boundary: 2025-03-15.
Publication age does not tell us whether a claim is still valid. Reading an old source again does not make its publication date newer. An update date does not establish that the passage we used was updated.
Read abstract, rationale, research questions, limitations and conclusion. A research position paper: reasoning traces may add monitoring value while remaining incomplete and potentially fragile. Authors' views are not necessarily their institutions' positions; cited experiments were not all independently reviewed.
Edition and machine-readable evidence
Content version 0.20.0. Evidence cutoff 2026-09-15; this does not mean every source was read on that day.
Pinned complete dataset · Complete evidence page · Agent consumption guide