llm-answers-question-gives-explanation-removing-input-concepts-explanation CLM-0205When an LLM answers a question and gives an explanation, removing from the input the concepts its explanation did not credit and re-querying on the reduced input raises measured explanation faithfulness over both standard prompting and prompts that ask for faithfulness, because the removal deletes the unmentioned influences rather than asking the model to report them.
Capability: Saying what actually drove the answer
Observed on
Tested on two datasets with question-plus-explanation prompts across multiple model families, using two faithfulness metrics; targets incompleteness (omitted influences), not unsou.
Sources
- Measured on two datasets, several model families, two independent faithfulness metrics, against standard prompting and faithfulness-encouraging prompting baselines. Only the abstract and introduction were available, so effect sizes, metric definitions, and how reliably concepts are extracted are not verifiable here; the design does not isolate whether gains come from removal or from the second query.
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- When an LLM gives an answer plus an explanation, deleting from the input those concepts the explanation does not credit and re-querying on the reduced input raises measured faithfulness over both standard prompting and prompts that ask for faithful explanations, because the re-query removes the unmentioned factors that could have influenced the original answer.Saying what actually drove the answer · reviewed by AI
- Unlearning methods that make a model answer "I don't know" about a forget set can leave the knowledge in place: the strongest IDK-style fine-tuning method still answered those same questions accurately when probed in a different format, and across nine unlearning methods on three model families none met a joint standard of utility and honesty on retained knowledge plus consistent, acknowledged forgetting; aligning representations after feature randomization (ReVa) roughly doubled the rejection rate that survives a second conversational turn.Retaining skills and safety after fine-tuning · reviewed by AI
- On factual QA (817-item TruthfulQA multiple-choice, plus a 300-item Natural Questions short-answer set), a prompt-only three-stage gate in which the model lists the information units a question needs, scores its support for each, and answers only when the mean score clears a threshold lowers the unconditional wrong-answer rate and raises accuracy on answered items for all eleven models tested, but most of the abstentions it produces fall on questions plain chain-of-thought answered correctly (about 69% at the primary operating point), so the gain is bought with lost coverage rather than with selective self-knowledge.Saying it does not know · reviewed by AI
- In multi-turn text-to-SQL where an underspecified question hides two dependent clarification layers, current LLM agents almost never surface both layers (joint identification under 5% for all 12 models tested), and this is not explained by how often they ask: ask rates spanning under 10% to over 60% all produce joint identification below 4%, and second-layer identification stays low (1.2-13.8%) even when the first layer is pre-resolved for them.Writing a reusable locator: regex, SQL, XPath · reviewed by AI
- In an open-ended LLM roleplay game where the character must withhold authored facts, gating generation through an external state tree — one retrieved permitted fact injected into the prompt, plus a separate verifier that decides whether the fact was actually conveyed before updating state — cuts the share of responses containing story-breaking unauthored content (17.8% to 6.3%) and stops premature reveal of the ending entirely, but makes some reveals appear unmotivated because the generator is forced to deliver the retrieved fact on that turn.Stating false facts confidently · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: low. Falsifier as drafted: Re-querying on the concept-reduced input yields no higher faithfulness than standard or faithfulness-encouraging prompting on the same datasets and metrics, or the gain disappears when the removal step is replaced by a no-op reduction. Proposed technique, not catalogued: Remove unmentioned concepts and re-query.