Factual errors in pretraining data cause language models to hold corresponding false beliefs.
Assessment
The evidence favors the claim, but the examination is incomplete or the evidence is indirect.
Read as a general causal tendency, the claim is well supported. Language models are trained to reproduce the statistics of their corpus. Controlled studies show that false facts injected into training data are then asserted by the model. The clearest is Alber et al. (Nature Medicine, 2025): replacing about 0.001% of training tokens with medical misinformation made models measurably more likely to produce medical errors, and standard benchmarks did not detect the change. Observational work points the same way: models reproduce common human misconceptions found in their pretraining data, as benchmarks such as TruthfulQA document.
The claim needs two qualifications. First, it is not true of every error. An error that appears once in a large corpus, among many correct statements of the same fact, is usually outweighed. Whether an error is learned depends on how often it appears, how consistent it is, and how often the correct version appears. Later training, such as fine-tuning or retrieval, can also suppress or override errors absorbed in pretraining. Second, calling these "beliefs" is a philosophical framing. The empirical finding is that models assert the falsehoods and behave as if they held them. Whether models have beliefs in a stronger sense is a separate dispute, and the causal finding does not depend on it.
Full reasoning: the evidence and decisions behind this verdict
Prior model: next-token training on a corpus should transmit consistent corpus errors, with frequency effects (long-tail facts are learned weakly, and errors are outvoted by correct mentions). Expected interventional evidence from poisoning and knowledge-injection studies. Search confirmed the anchor study: Alber et al., "Medical large language models are vulnerable to data-poisoning attacks," Nature Medicine 31:618–626 (2025), www.nature.com/articles/s41591-024-03445-1. Replacing 0.001% of training tokens with medical misinformation produced models more likely to propagate medical errors, and standard medical benchmarks did not detect it. That is direct causal evidence that pretraining errors produce matching false outputs. The observational subclaim about reproducing misconceptions (TruthfulQA and similar) supports the claim but cannot by itself separate the training-data cause from other causes. Scope limits come from my own reasoning plus knowledge of the frequency and long-tail literature (e.g., Kandpal et al. 2023), which I did not re-verify in this pass. The unqualified universal reading ("any factual error") is false. The generic causal reading is well supported, and the credence of 0.85 applies to that reading. The word "belief" is treated as shorthand for behavioral disposition. What would change this: evidence that injected falsehoods are reliably suppressed at realistic frequencies, or that post-training removes them. Budget limited this pass to a light verification.
Decomposition
The claims this one rests on directly. ↗︎ opens a subclaim; the map shows how they fit together.
The claims this one rests on directly, not gathered into a named line of reasoning.
Because controlled experiments that inject false facts into training data find that models then assert them, and because models reproduce common human misconceptions present in their pretraining data, errors in pretraining data causally shape what models state as fact. Since next-token training rewards matching the corpus, a falsehood the corpus states consistently is learned like any other fact.
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.
This is the causal version of the claim that language models reproduce misconceptions found in their training data. Correlation between corpus errors and model errors does not by itself establish causation, but controlled evidence is available in principle: studies that inject false facts into training or fine-tuning data and measure whether models then assert them, and work on data poisoning and memorization. The claim is stronger and harder to establish than the observational version, and its scope (all factual errors, or only frequently repeated ones) should be settled in assessment.
The proposal states a single reusable proposition about the world: that factual errors in pre-training corpora causally produce corresponding false beliefs in language models. It can be assessed with evidence, such as data-poisoning studies, fact-injection fine-tuning experiments, and memorization work. The contributor correctly separates it from the observational claim that models reproduce training-data misconceptions, because the causal claim turns on different considerations (controlled interventions, not correlation). The wording is close to a workable canonical form. The Steward can settle questions of scope, such as all errors versus frequently repeated ones, and of what counts as a "belief". The submission concerns no private person and is in good faith. The Matcher will decide whether it already exists under another wording.
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 2, 2026. Every judgment on this page is accompanied by a reasoning trace.