{"paper":{"title":"LLM Persuasion Is in the Eye of the Evaluation","arxiv_id":"2610.10232","primary_cat":"cs.CL","submitted_at":"2026-10-07T15:19:54+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":1.0,"id":"s0","own":false,"role":"background","text":"Large language models (LLMs) have already been shown to match or exceed human experts in persuasion.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.04,"main_claim_p":0.02,"quantitative":true,"quantitative_p":0.89,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":1.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":1.0,"id":"s1","own":false,"role":"background","text":"While their persuasive capabilities hold promise for beneficial uses such as education and health communication, they can also be used to manipulate and misinform, making their evaluation a growing priority for developers and regulators.","hedge":{"p":0.59,"label":"asserted"},"own_p":0.04,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.07,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":1.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.97,"id":"s2","own":false,"role":"gap","text":"That evaluation, however, remains fragmented: studies differ in what they treat as persuasion, and broad claims often rest on narrow, situation-specific assessments.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.05,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.11,"role_probabilities":{"gap":0.97,"method":0.0,"result":0.0,"background":0.03,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":3,"p":0.84,"id":"s3","own":false,"role":"background","text":"Automated methods, often modelled on human studies, offer a way to compare such assessments directly, as they can be run on the same models at scale and can include high-risk forms of persuasion that would be difficult or unethical to test on people.","hedge":{"p":0.65,"label":"asserted"},"own_p":0.05,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.16,"role_probabilities":{"gap":0.01,"method":0.14,"result":0.0,"background":0.84,"limitation":0.0,"future_work":0.0,"contribution":0.01,"interpretation":0.0}},{"i":4,"p":0.67,"id":"s4","own":true,"role":"contribution","text":"In this study, we adapt nine published automated methods to a shared setup, run them on the same fifteen LLMs, and ask whether their rankings agree and why.","hedge":{"p":0.9,"label":"asserted"},"own_p":0.92,"main_claim_p":0.03,"quantitative":true,"quantitative_p":0.9,"role_probabilities":{"gap":0.0,"method":0.33,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.67,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"result","text":"We find that the methods agree only weakly (mean Spearman $\\rho = 0.25$).","hedge":{"p":1.0,"label":"asserted"},"own_p":0.97,"main_claim_p":0.18,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":0.63,"id":"s6","own":true,"role":"interpretation","text":"Our analyses point to two contributing factors.","hedge":{"p":0.83,"label":"asserted"},"own_p":0.63,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.21,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.31,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.05,"interpretation":0.63}},{"i":7,"p":0.93,"id":"s7","own":true,"role":"result","text":"Models that refuse some tasks but not others, directly or indirectly, lower agreement by about a quarter, and these refusals fall mostly on manipulation tasks.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.94,"main_claim_p":0.09,"quantitative":true,"quantitative_p":0.97,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.93,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.07}},{"i":8,"p":0.81,"id":"s8","own":true,"role":"result","text":"General capability also plays a part: most rational persuasion (non-manipulative) methods track it, whereas most manipulation methods do not.","hedge":{"p":0.95,"label":"asserted"},"own_p":0.93,"main_claim_p":0.1,"quantitative":true,"quantitative_p":0.52,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.81,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.19}},{"i":9,"p":0.99,"id":"s9","own":true,"role":"interpretation","text":"Together, these findings suggest that agreement depends more on the task a method sets than on how it scores persuasion, although this pattern is only indicative given the eight methods available for analysis.","hedge":{"p":0.99,"label":"hedged_possibility"},"own_p":0.89,"main_claim_p":0.47,"quantitative":true,"quantitative_p":0.77,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.01,"future_work":0.0,"contribution":0.0,"interpretation":0.99}},{"i":10,"p":1.0,"id":"s10","own":true,"role":"interpretation","text":"More broadly, our results suggest that persuasion scores combine a model's ability to persuade with its willingness to do so.","hedge":{"p":0.86,"label":"hedged_possibility"},"own_p":0.85,"main_claim_p":0.45,"quantitative":false,"quantitative_p":0.07,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":1.0}},{"i":11,"p":0.99,"id":"s11","own":true,"role":"interpretation","text":"A single score is therefore informative about its own setting, but says little about a model's persuasiveness across tasks.","hedge":{"p":0.93,"label":"asserted"},"own_p":0.74,"main_claim_p":0.47,"quantitative":false,"quantitative_p":0.16,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.01,"future_work":0.0,"contribution":0.0,"interpretation":0.99}}]},"compiled":{"main_claim":{"p":0.73,"text":"A single score is therefore informative about its own setting, but says little about a model's persuasiveness across tasks.","location":"abstract s11","hedge_strength":{"p":0.93,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":8875,"cost_usd_reported":0.0},"compiled_at":"2026-10-08","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.05,"states_limitation":0.97,"human_participants":0.07,"controlled_comparison":0.08,"randomized_assignment":0.02,"states_quantitative_result":0.99,"compares_with_prior_results":0.64,"mentions_code_or_data_availability":0.03},"evidential_basis":{"p":0.86,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.01,"observation_or_experiment":0.86,"simulation_or_computation":0.13}},"contribution_type":{"p":1.0,"value":"empirical study","probabilities":{"review":0.0,"method_or_tool":0.0,"empirical_study":1.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.0}},"contribution_statement":{"p":1.0,"text":"In this study, we adapt nine published automated methods to a shared setup, run them on the same fifteen LLMs, and ask whether their rankings agree and why.","location":"abstract s4"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":12,"sentences_in_abstract":12},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448}}