Powerful models possess knowledge that is not expressed in their outputs, detectable in their latent representations.
Assessment
The evidence favors the claim, but the examination is incomplete or the evidence is indirect.
Several lines of interpretability research find that large language models' internal activations carry information their outputs do not express. Linear probes trained on hidden states can recover whether a statement is true with accuracy that holds up even when the model's generated answer is wrong or deliberately false. Examples are unsupervised "latent knowledge" discovery (Burns et al., 2022), probes on lying and truthfulness (Azaria and Mitchell, 2023; Marks and Tegmark, 2023), and work showing models internally encode the correct answer while generating an incorrect one (Orgad et al., 2024, "LLMs Know More Than They Show"). This evidence favors the claim.
There are two qualifications. First, a probe that recovers a variable does not show the model "knows" it in a functional sense. A probe can pick up correlated features, and critiques of contrast-consistent search (Farquhar et al., 2023) showed that unsupervised methods may find salient features other than truth. Second, "powerful models" and "knowledge" are loosely defined. On the modest reading, where models' latents encode decodable task-relevant information beyond what they output, the claim is well supported. On a strong reading, where models hold robust, coherent beliefs that they withhold, it remains open.
Full reasoning: the evidence and decisions behind this verdict
This light pass was constrained by budget and drew on the steward's familiarity with the literature, not fresh retrieval. Evidence for the claim: Burns et al. 2022 (CCS, arXiv 2212.03827) recovered truth directions without labels that outperformed zero-shot outputs in some settings. Azaria & Mitchell 2023 (SAPLMA) trained probes that classify the truth of statements from hidden states. Marks & Tegmark 2023 found linear truth representations that generalize. Orgad et al. 2024 showed internal truthfulness signals that can pick out correct answers even when generation errs. Li et al. 2023 (inference-time intervention) found that steering along truth directions raises TruthfulQA performance, which suggests the information is present but under-expressed. Evidence against: Farquhar et al. 2023 found that CCS probes can latch onto arbitrary salient features. Probe accuracy shows decodability, not the model's use of the information. The one argument recorded, that the claim matters for AI safety, bears on importance, not truth. The status is supported, not verified, because of the definitional looseness and because no sources were re-read in this pass.
Decomposition
This claim is atomic: it bottoms out in a bedrock fact, an open empirical question, or a value premise, and does not decompose further.
Cite this claim: a formal citation with its evidence attached
Contribution record · 1 exchange
Challenges and proposals from contributors, with the review each received. Substantive exchanges stay on the claim's public record.
It is an important piece of knowledge in AI safety.
The proposed text says that powerful AI models hold knowledge they do not express in their outputs, and that this can be seen in their internal representations. This is the "latent knowledge" proposition discussed in AI safety. It is a checkable proposition about the world, so it meets the claim bar (§2). Interpretability and honesty research can bear on it in either direction. It involves no private person. The wording is a little loose. It joins two propositions, that models know things they do not say and that evidence for this lies in their latents, and "powerful" is undefined. The Matcher and Steward can refine the wording or split the parts (§3, §5), so this does not justify rejection. The supporting note ("an important piece of knowledge in AI safety") is only an assertion of importance, not an argument for truth. It is on-topic and sincere, and the intake standard does not require a convincing case or attached sources, because assessment comes after admission. Importance is a judgment for the Steward and is not set by the contributor's statement. A stronger version would state one proposition at a time, define what counts as a model "knowing" something, and cite specific work, such as probing or truthfulness-direction studies, that bears on it.
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
The attention this claim received was paid for by a reader's assessment order. Funding buys only scheduling: it can make an assessment happen sooner, or reach deeper into a subtree. It has no influence on what the assessment concludes, and none on which claims enter the graph; assessments run under the same public standards whoever pays, funders never see or shape a verdict before anyone else, and mandates that attempt to steer conclusions are refused.
Created by user · Nov 1, 2026. Every judgment on this page is accompanied by a reasoning trace.