{"id":"https://openalex.org/W7204769485","doi":"https://doi.org/10.48550/arxiv.2608.27754","title":"Efficient Auto-Interpretability of AI Models in Biology","display_name":"Efficient Auto-Interpretability of AI Models in Biology","publication_year":2026,"publication_date":"2026-08-27","ids":{"openalex":"https://openalex.org/W7204769485","doi":"https://doi.org/10.48550/arxiv.2608.27754"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.27754","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.27754","pdf_url":"https://arxiv.org/pdf/2608.27754","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.27754","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5028735308","display_name":"Piotr Jedryszek","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Jedryszek, Piotr","raw_affiliation_strings":["Department of Biology , University of Oxford , Oxford , UK","Evolvere Biosciences , London , UK"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Biology , University of Oxford , Oxford , UK","institution_ids":["https://openalex.org/I40120149"]},{"raw_affiliation_string":"Evolvere Biosciences , London , UK","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5044004993","display_name":"Oliver M. Crook","orcid":"https://orcid.org/0000-0001-5669-8506","observed_orcids":["https://orcid.org/0000-0001-5669-8506"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Crook, Oliver M.","raw_affiliation_strings":["Department of Chemistry , University of Oxford , Oxford , UK","Kavli Institute for Nanoscience Discovery , University of Oxford , Oxford , UK"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Chemistry , University of Oxford , Oxford , UK","institution_ids":["https://openalex.org/I40120149"]},{"raw_affiliation_string":"Kavli Institute for Nanoscience Discovery , University of Oxford , Oxford , UK","institution_ids":["https://openalex.org/I40120149"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I40120149"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9984999895095825,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9984999895095825,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12254","display_name":"Machine Learning in Bioinformatics","score":0.0003000000142492354,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T10211","display_name":"Computational Drug Discovery Methods","score":9.999999747378752e-05,"subfield":{"id":"https://openalex.org/subfields/3002","display_name":"Drug Discovery"},"field":{"id":"https://openalex.org/fields/30","display_name":"Pharmacology, Toxicology and Pharmaceutics"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/sparse-autoencoders","display_name":"sparse autoencoders","score":0.9100000262260437},{"id":"https://openalex.org/keywords/interpretability","display_name":"interpretability","score":0.6919999718666077},{"id":"https://openalex.org/keywords/latent-space-exploration","display_name":"latent space exploration","score":0.46700000762939453},{"id":"https://openalex.org/keywords/intruder-detection","display_name":"intruder detection","score":0.9710000157356262},{"id":"https://openalex.org/keywords/motif-discovery","display_name":"motif discovery","score":0.4050000011920929},{"id":"https://openalex.org/keywords/testable-predictions","display_name":"falsifiable predictions","score":0.906000018119812},{"id":"https://openalex.org/keywords/protein-structure","display_name":"protein structure","score":0.3930000066757202},{"id":"https://openalex.org/keywords/computational-cost-reduction","display_name":"computational complexity reduction","score":0.5210000276565552}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.9721999764442444},{"id":"https://openalex.org/C116222747","wikidata":"https://www.wikidata.org/wiki/Q220888","display_name":"Falsifiability","level":2,"score":0.7258999943733215},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6703000068664551},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6373000144958496},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6182000041007996},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.6175000071525574},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.5938000082969666},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5623000264167786},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.42800000309944153},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.27799999713897705},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2547000050544739}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.27754","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.27754","pdf_url":"https://arxiv.org/pdf/2608.27754","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.27754","doi":"https://doi.org/10.48550/arxiv.2608.27754","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.27754","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.27754","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.27754","pdf_url":"https://arxiv.org/pdf/2608.27754","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204769485.pdf","grobid_xml":"https://content.openalex.org/works/W7204769485.grobid-xml"},"referenced_works_count":10,"referenced_works":["https://openalex.org/W4396721167","https://openalex.org/W4404565424","https://openalex.org/W4404691028","https://openalex.org/W4406950770","https://openalex.org/W4416102053","https://openalex.org/W4417275958","https://openalex.org/W7118185881","https://openalex.org/W7155417360","https://openalex.org/W7159661641","https://openalex.org/W7202368571"],"related_works":["https://openalex.org/W7171672762","https://openalex.org/W7134118817","https://openalex.org/W7171621853","https://openalex.org/W7043438366","https://openalex.org/W7170187141","https://openalex.org/W7165630099","https://openalex.org/W4417275958","https://openalex.org/W7161630786","https://openalex.org/W7138031500","https://openalex.org/W4413312092"],"abstract_inverted_index":{"Sparse":[0],"autoencoders":[1],"(SAEs),":[2],"and":[3,13,49,68,143,156],"other":[4,14],"interpretability":[5],"methods":[6],"could":[7],"turn":[8],"AI":[9],"models":[10],"in":[11,122],"Biology":[12],"fields":[15],"into":[16,64,115],"engines":[17],"of":[18,26,154,188],"scientific":[19],"discovery":[20],"by":[21],"explaining":[22],"the":[23,70,126,157,161,178],"superhuman":[24],"capabilities":[25],"those":[27],"models.":[28],"However,":[29],"a":[30,65,95,100,104,108,175],"latent":[31,140],"is":[32,42],"only":[33],"useful":[34],"if":[35],"we":[36,113],"know":[37],"three":[38],"things:":[39],"whether":[40,44,50,94],"it":[41,45],"coherent,":[43],"can":[46,119],"be":[47,120,183],"described,":[48],"that":[51],"description":[52,111],"has":[53],"predictive":[54],"power.":[55],"These":[56],"questions":[57],"are":[58,83,164],"routinely":[59],"conflated.":[60],"We":[61],"assemble":[62],"them":[63],"single":[66],"pipeline":[67],"report":[69],"practical":[71],"innovations":[72],"each":[73],"stage":[74],"required.":[75],"First,":[76],"cross-seed":[77],"dictionary":[78],"stability":[79,130,181],"prioritises":[80],"which":[81,112,118],"latents":[82,96,134],"worth":[84],"spending":[85],"resources":[86],"to":[87],"investigate.":[88],"Second,":[89],"an":[90],"intruder-detection":[91],"task":[92],"asks":[93],"activating":[97],"examples":[98],"share":[99],"recognizable":[101],"pattern.":[102],"Third,":[103],"separate":[105],"pass":[106],"proposes":[107],"candidate":[109],"biological":[110],"convert":[114],"falsifiable":[116],"predictions":[117],"tested":[121],"silico.":[123],"Deployed":[124],"on":[125],"Boltz-1":[127],"Pairformer":[128],"trunk,":[129],"prioritisation":[131],"finds":[132],"interpretable":[133],"using":[135],"about":[136],"4.4":[137],"times":[138,146],"fewer":[139],"evaluations":[141],"each,":[142],"at":[144],"5.2":[145],"lower":[147],"measured":[148],"cost,":[149],"while":[150],"recovering":[151],"over":[152],"half":[153],"them,":[155],"external":[158],"check":[159],"shows":[160],"surfaced":[162],"motifs":[163],"significantly":[165],"enriched":[166],"for":[167,185],"their":[168],"claimed":[169],"annotations.":[170],"The":[171],"results":[172],"also":[173],"suggest":[174],"possible":[176],"tension:":[177],"cross-":[179],"seed":[180],"might":[182],"selecting":[184],"some":[186],"types":[187],"features,":[189],"like":[190],"structure-related":[191],"ones,":[192],"much":[193],"more":[194],"than":[195],"others,":[196],"such":[197],"as":[198],"function-related":[199],"features.":[200]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-01T00:00:00"}