Candidates

Papers that do not have a page yet: 18 accepted and waiting to be written, and 196 one citation away from the wiki.

Stubs (18)

Papers accepted into the wiki whose pages are still to be written. Each has its bibliographic details and a one-line description so far.

  • Looking Inward: Language Models Can Learn About Themselves by Introspection Binder et al. (2024) · coreA model fine-tuned to predict properties of its own answers does so more accurately than a second model fine-tuned on the same data about it, and its predictions follow its behavior when that behavior is changed. The effect appears only on simple tasks and does not transfer to other self-knowledge tasks.
  • Can Language Models Explain Their Own Classification Behavior? Sherburn et al. (2024) · coreModels that classify text by a simple rule often cannot state that rule. GPT-3 fails in free text even after fine-tuning on correct explanations, GPT-4 succeeds 72% of the time on the rules it classifies best, and the authors say a correct statement would still not show that it came from introspection.
  • Tell me about yourself: LLMs are aware of their learned behaviors Betley et al. (2025) · coreModels fine-tuned to follow a policy that the training data never describes, such as taking risky gambles or writing insecure code, can describe that policy when asked, with no examples in the prompt. Backdoored models can sometimes say they have a backdoor, but do not state its trigger in free text unless trained on reversed examples.
  • Does It Make Sense to Speak of Introspection in Large Language Models? Comsa & Shanahan (2025) · coreProposes that an LLM self-report is introspective if it accurately describes an internal state through a causal process linking that state to the report. On that definition, the authors argue, Gemini's account of how it wrote a poem is not introspection, but its inference of its own sampling temperature from text it has just written is a minimal case.
  • Training Language Models to Explain Their Own Computations Li et al. (2025) · coreFine-tuned to put the results of interpretability procedures into words, a model explains its own features and intervention outcomes more accurately than a different model trained on the same examples, even a larger one. It also needs far less training data.
  • Emergent Introspective Awareness in Large Language Models Lindsey (2025) · coreClaude Opus 4 and 4.1 detect and correctly name a concept vector injected into their activations on about 20% of trials at the best layer and strength, with no false positives on control trials. Some models also consult their earlier activations to judge whether a prefilled output was their own, but the author calls these abilities highly unreliable and context-dependent.
  • Tests of LLM introspection need to rule out causal bypassing Morris & Plunkett (2025) · coreAn intervention that changes a model's internal state can also cause an accurate report of that state by a path that skips the state, so accuracy after an intervention does not show the report is grounded. The authors name this causal bypassing and say the only test they know that rules it out is asking a model whether a concept was injected, a claim a later edit to the post hedges.
  • Self-Interpretability: LLMs Can Describe Complex Internal Processes that Drive Their Decisions, and Improve with Training Plunkett et al. (2025) · coreAfter fine-tuning on choices generated from random attribute weights, GPT-4o and GPT-4o-mini state those weights with a correlation of about 0.5 to the weights their choices reveal. Training on correct reports raises this to about 0.75 on held-out decisions and also improves reports about preferences that were never fine-tuned.
  • Language Models Fail to Introspect About Their Knowledge of Language Song et al. (2025) · coreAcross 21 open-source models, answers to metalinguistic prompts predict a model's own string probabilities no better than they predict those of a near-identical model. The authors find no evidence of privileged self-access to grammatical knowledge or word predictions.
  • Privileged Self-Access Matters for Introspection in AI Song et al. (2025) · coreProposes that introspection in AI be defined by privileged self-access: a process that tells a model about its internal states more reliably than any process of equal or lower computational cost available to a third party. In a temperature self-report task, four models' reports follow the framing of the prompt, and self-reflection is no more accurate than another model's prediction, with accuracy no better than a random baseline.
  • Detecting the Disturbance: A Nuanced View of Introspective Abilities in LLMs Hahami et al. (2026) · coreIn Llama 3.1 8B, apparent success at answering "did you detect an injected thought?" is fully explained by the injection pushing the model toward "yes" on any question. The same model can still say which of ten sentences was injected (up to 88%) and which of two injections was stronger (up to 83%), but only when the injection is in the first few layers.
  • Latent Introspection: Models Can Detect Prior Concept Injections Pearson-Vogel et al. (2026) · coreQwen2.5-Coder-32B carries information about a concept vector that was injected during an earlier turn and then removed, including which concept it was. The signal peaks around layers 58 to 62, is weakened by the final layers, and reaches the output only under some prompts: with a document explaining introspection, P("yes") is 39.9% with injection and 0.8% without.
  • Taken out of context: On measuring situational awareness in LLMs Berglund et al. (2023) · adjacentModels fine-tuned on written descriptions of fictitious chatbots, with no examples, can sometimes act as described when the description is absent from the prompt, but only if each description is paraphrased many times; accuracy rises with model size. The paper proposes this out-of-context reasoning as a measurable component of situational awareness.
  • Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data Treutlein et al. (2024) · adjacentA model fine-tuned on documents that each hold one indirect observation of a hidden fact (a distance to an unknown city, a coin flip, one input-output pair of a function) can afterwards state the fact and use it, with no examples in the prompt and no chain of thought. This beat in-context learning on the paper's five tasks but was unreliable.
  • Explicitly unbiased large language models still form biased associations Bai et al. (2025) · adjacentEight chat models that pass standard bias benchmarks still pair social groups with stereotyped words, and make matching choices between people, when tested with indirect prompts adapted from psychology. The models are never asked about themselves.
  • Eliciting Secret Knowledge from Language Models Cywiński et al. (2025) · adjacentModels fine-tuned to act on a secret while denying they know it can still be made to give it up: prefill attacks let an auditor recover the secret with over 90% success in two of three settings. Logit-lens and sparse-autoencoder readouts of the activations also help the auditor, though less.
  • On the Biology of a Large Language Model Lindsey et al. (2025) · adjacentCircuit tracing in Claude 3.5 Haiku finds the model's account of its own computation matching the mechanism in one case and diverging in others: it describes carry-the-one addition while computing the sum another way, and a chain of thought can be genuine, invented, or worked backwards from a user's hint. Whether it answers a question or says it does not know depends on "known answer" features that can be active for a familiar name when the answer is not known.
  • Simple Mechanistic Explanations for Out-Of-Context Reasoning Wang et al. (2025) · adjacentOn Gemma 3 12B, a one-layer LoRA fine-tune that produces out-of-context reasoning mostly adds a single constant vector. A steering vector trained directly on the same data also makes the model state a behavior or fact it was only trained to act on.

One citation away (196)

Papers that have not been accepted. The crawler looked at the references and citers of every paper page and saw 1159 distinct neighbors. A neighbor is listed here if it connects to at least 2 wiki papers, or if someone added it as a lead. The triage labels are suggestions, made from each candidate's title and its place in the citation graph and not from reading it. A candidate becomes a page only after a person accepts it.

Last crawled 2026-10-07. Source: Semantic Scholar Graph API, with arXiv HTML bibliographies where Semantic Scholar has no reference list. Also as JSON.

Suggested: add (43)

Suggested: maybe (52)

Suggested: skip (101)