English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Large language models (LLMs) are trained to imitate humans to explain humandecisions. However, do LLMs explain themselves? Can they help humans buildmental models of how LLMs process different inputs? To answer these questions,we propose to evaluate $\textbf{counterfactual simulatability}$ of naturallanguage explanations: whether an explanation can enable humans to preciselyinfer the model's outputs on diverse counterfactuals of the explained input.For example, if a model answers "yes" to the input question "Can eagles fly?"with the explanation "all birds can fly", then humans would infer from theexplanation that it would also answer "yes" to the counterfactual input "Canpenguins fly?". If the explanation is precise, then the model's answer shouldmatch humans' expectations.  We implemented two metrics based on counterfactual simulatability: precisionand generality. We generated diverse counterfactuals automatically using LLMs.We then used these metrics to evaluate state-of-the-art LLMs (e.g., GPT-4) ontwo tasks: multi-hop factual reasoning and reward modeling. We found that LLM'sexplanations have low precision and that precision does not correlate withplausibility. Therefore, naively optimizing human approvals (e.g., RLHF) maynot be a sufficient solution.

Do Models Explain Themselves? Counterfactual Simulatability of Natural Language Explanations