{"paper":{"title":"Discovery Foundation Models: Toward Open-Ended Discovery Intelligence","arxiv_id":"2609.15973","primary_cat":"cs.CL","submitted_at":"2026-09-14T17:56:00+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":"Foundation models have progressed from learning and reasoning over existing knowledge, to increasingly learning through action, tool use, and outcome feedback.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.07,"main_claim_p":0.04,"quantitative":false,"quantitative_p":0.19,"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":0.37,"id":"s1","own":false,"role":"contribution","text":"We argue that the next frontier is a further transition: from solving and acting within problems specified by humans to participating in the process by which new problems, representations, explanations, and knowledge are created.","hedge":{"p":0.77,"label":"asserted"},"own_p":0.13,"main_claim_p":0.31,"quantitative":false,"quantitative_p":0.17,"role_probabilities":{"gap":0.35,"method":0.0,"result":0.0,"background":0.1,"limitation":0.0,"future_work":0.16,"contribution":0.37,"interpretation":0.02}},{"i":2,"p":0.83,"id":"s2","own":true,"role":"contribution","text":"We refer to this capability as Discovery Intelligence.","hedge":{"p":0.95,"label":"asserted"},"own_p":0.57,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.01,"method":0.03,"result":0.0,"background":0.07,"limitation":0.0,"future_work":0.0,"contribution":0.83,"interpretation":0.06}},{"i":3,"p":0.83,"id":"s3","own":true,"role":"contribution","text":"We formulate Discovery Foundation Models (DFMs) as general-purpose model systems for open-ended discovery.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.76,"main_claim_p":0.09,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.17,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.83,"interpretation":0.0}},{"i":4,"p":0.51,"id":"s4","own":true,"role":"contribution","text":"A DFM operates over a revisable research state and supports seven coupled capabilities spanning problem discovery, formulation, representation construction, hypothesis formation, intervention, evidence-grounded revision, and continual discovery improvement.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.62,"main_claim_p":0.09,"quantitative":true,"quantitative_p":0.95,"role_probabilities":{"gap":0.0,"method":0.49,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.51,"interpretation":0.0}},{"i":5,"p":0.53,"id":"s5","own":true,"role":"contribution","text":"We instantiate this framework with Zetema, which couples explicit research-state dynamics, verification and experimental gating, external grounding, and cross-task Discovery Skill evolution.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.88,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.06,"role_probabilities":{"gap":0.0,"method":0.47,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.53,"interpretation":0.0}},{"i":6,"p":0.64,"id":"s6","own":true,"role":"contribution","text":"We further ground the framework with GALILEO, a real therapeutic-discovery system in which Dry-Lab reasoning, robotic and hands-on Wet-Lab experimentation, external biological evidence, and iterative hypothesis and design revision form a closed physical discovery loop.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.67,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.08,"role_probabilities":{"gap":0.0,"method":0.36,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.64,"interpretation":0.0}},{"i":7,"p":0.75,"id":"s7","own":true,"role":"contribution","text":"We then formulate a unified approach to capability formation and process-centered evaluation, enabling discovery behavior to be trained, improved, and measured beyond final-answer performance.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.85,"main_claim_p":0.13,"quantitative":false,"quantitative_p":0.06,"role_probabilities":{"gap":0.0,"method":0.25,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.75,"interpretation":0.0}},{"i":8,"p":0.54,"id":"s8","own":true,"role":"interpretation","text":"Together, these components establish discovery as a learnable, executable, and evaluable capability of foundation-model systems.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.67,"main_claim_p":0.62,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.08,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.38,"interpretation":0.54}},{"i":9,"p":1.0,"id":"s9","own":false,"role":"interpretation","text":"We view this shift as a broader progression in intelligence scaling: from learning over existing knowledge, to learning from action outcomes, and ultimately to participating in the construction, testing, and revision of the structures through which new knowledge is discovered.","hedge":{"p":0.83,"label":"asserted"},"own_p":0.13,"main_claim_p":0.34,"quantitative":false,"quantitative_p":0.32,"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":10,"p":0.45,"id":"s10","own":false,"role":"future_work","text":"Code: https://github.com/Gen-Verse/DFM-Plans","hedge":{"p":0.76,"label":"unclear"},"own_p":0.37,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.12,"result":0.01,"background":0.03,"limitation":0.0,"future_work":0.45,"contribution":0.38,"interpretation":0.01}}]},"compiled":{"main_claim":{"p":0.97,"text":"Together, these components establish discovery as a learnable, executable, and evaluable capability of foundation-model systems.","location":"abstract s8","hedge_strength":{"p":1.0,"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":8234,"cost_usd_reported":0.0},"compiled_at":"2026-09-23","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.08,"states_limitation":0.07,"human_participants":0.1,"controlled_comparison":0.02,"randomized_assignment":0.01,"states_quantitative_result":0.08,"compares_with_prior_results":0.1,"mentions_code_or_data_availability":0.94},"evidential_basis":{"p":0.92,"value":"argument","probabilities":{"argument":0.92,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":0.06,"simulation_or_computation":0.02}},"contribution_type":{"p":0.56,"value":"position or commentary","probabilities":{"review":0.0,"method_or_tool":0.44,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.56}},"contribution_statement":{"p":0.63,"text":"We formulate Discovery Foundation Models (DFMs) as general-purpose model systems for open-ended discovery.","location":"abstract s3"}},"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":11,"sentences_in_abstract":11},"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,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}