[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-661261-105":59,"doc-detail-661261-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","understanding-machine-taste-aesthetic-agreement-across-foundation-models-in-k-pop","Understanding Machine Taste - Aesthetic Agreement Across Foundation Models in K-pop","","As foundation models move into creative industries, identifying their aesthetic preferences supports better AI-assisted creative choices. This study examines how models evaluate K-pop songs and music videos, framing aesthetic evaluation as pairwise ranking. The work measures pairwise inter-model agreement, group-level consensus, and crossmodal preference alignment, showing stronger agreement and consensus in visual evaluation than audio, with only weak audio-visual alignment. Visual consensus is also linked to real-world popularity using YouTube history, yielding positive but varying correlations by artist group.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/understanding-machine-taste-aesthetic-agreement-across-foundation-models-in-k-pop/661261/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/understanding-machine-taste-aesthetic-agreement-across-foundation-models-in-k-pop/661261.png","ImageObject",300,407,{"name":92,"@type":93},"Mabel","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-29","2026-10-28",true,{"@type":102,"interactionType":103,"userInteractionCount":8},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"How does the paper define and measure “machine taste” for foundation models?","Question",{"text":112,"@type":113},"It models aesthetic evaluation as a pairwise ranking problem on K-pop songs (audio) and music videos (visual), then measures agreement between models and consensus across artist groups.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What does the study find about inter-model agreement in audio versus visual evaluation?",{"text":117,"@type":113},"The results show stronger pairwise inter-model agreement and higher group-level consensus in visual evaluation than in audio evaluation.",{"name":119,"@type":110,"acceptedAnswer":120},"How are visual preference rankings related to real-world popularity?",{"text":121,"@type":113},"The paper compares model-derived visual consensus rankings with historical YouTube viewership data, finding positive but varying correlations across artist groups.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},661261,1793161806,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":8,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":129,"read_time":144},7971474920318,"https://ap-avatar.wpscdn.com/avatar/74000ee537e6b0ca360?x-image-process=image/resize,m_fixed,w_180,h_180&k=1788430171632181674","Understanding Machine Taste: Aesthetic Agreement Across Foundation  \nModels in K-pop  \nKaishen Wang, Yucheng Pan, Xuehao Cui, Hanling Tian, Heng Huang  \nUniversity of Maryland, College Park {kaishen, ycpan, cedrus, hanling, [heng}@umd.edu](heng}@umd.edu)  \narXiv : 26 10 . 1773 1v 1 [cs .CL] 13 Oct 2026  \nAbstract  \nAs foundation models enter creative industries, understanding their aesthetic preferences becomes an important step toward AI-assisted creative decision making. In this work, we investigate how foundation models evaluate creative content and characterize their “machine taste” using K-pop songs and music videos as a testbed. We formulate aesthetic evaluation as a pairwise ranking problem and systematically examine pairwise inter-model agreement, group-level consensus, and crossmodal preference alignment. Our results reveal stronger pairwise inter-model agreement and group-level consensus in visual evaluation than in audio evaluation, while preferences across the two modalities remain weakly aligned. We further examine how visual consensus rankings relate to real-world popularity using historical YouTube viewership data, finding positive but varying correlations across artist groups. These findings provide initial insights into machine aesthetic preferences and lay the groundwork for future AI-assisted creative decision making that considers creative content, artist identity, audience preferences, and market dynamics.  \n1 Introduction  \nRecent foundation models (Ma et al., 2024 ; Yang et al., 2025 ; Xu et al., 2025 ; Yang et al., 2026 ; Ghosh et al., 2026 ; Ye et al., 2026 ; Wang et al., 2026) have become increasingly capable across language, vision, and audio, and are gradually being integrated into real-world workflows beyond conventional benchmark-oriented tasks. This transition is particularly important in domains where decisions are not determined by a single correct answer. In creative industries, for example, choices such as selecting a song, developing a visual concept, or evaluating creative content are inherently subjective and often depend on aesthetic judgment rather than objective correctness (Zhao et al., 2025 ; Fein et al., 2026 ; Chen et al., 2026) .  \nThe entertainment industry provides a timely example of this transition, as major companies are increasingly exploring AI applications in their creative and production workflows. SM Entertainment (SM Entertainment, 2026) has identified AI asa key component of its SM NEXT 3.0 strategy, aiming to improve production efficiency and support more sophisticated fan-and market-informed decision making. HYBE (HYBE, 2026) has expanded its AI capabilities through Supertone, which develops AI-based audio technologies for generating realistic singing and acting voices. JYP Entertainment (JYP Entertainment, 2025) has similarly expanded its AI-related initiatives, including plans to develop an AI artist as part of its future content portfolio. These developments reflect the growing role of AI in creative industries and point toward broader opportunities for AI-assisted creative and strategic decision making. An important first step toward this vision is to understand how foundation models themselves evaluate creative content and what aesthetic preferences they exhibit.  \nUnlike factual question answering or mathematical reasoning, aesthetic evaluation does not admit a unique ground-truth answer (Marco et al., 2025) . While prior studies have investigated the aesthetic evaluation capabilities and biases of foundation models (Li et al., 2024 ; Yao and Zhuang, 2025 ; Li et al., 2025), understanding how these models evaluate creative content remains challenging without a universally accepted reference for aesthetic quality. In this context, examining aesthetic preference agreement provides a useful starting point for investigating machine taste, allowing us to examine whether shared preferences emerge between individual models, among models evaluating the same artist group","cbCaigNE19YtArRx","https://ap.wps.com/l/cbCaigNE19YtArRx","pdf",1587149,17,"English","# Introduction\n## Creative Domain: K-pop\n## Study Overview and Research Questions\n## Methods: Pairwise Ranking and Agreement Metrics\n## Results and Popularity Analysis","[{\"question\":\"How does the paper define and measure “machine taste” for foundation models?\",\"answer\":\"It models aesthetic evaluation as a pairwise ranking problem on K-pop songs (audio) and music videos (visual), then measures agreement between models and consensus across artist groups.\"},{\"question\":\"What does the study find about inter-model agreement in audio versus visual evaluation?\",\"answer\":\"The results show stronger pairwise inter-model agreement and higher group-level consensus in visual evaluation than in audio evaluation.\"},{\"question\":\"How are visual preference rankings related to real-world popularity?\",\"answer\":\"The paper compares model-derived visual consensus rankings with historical YouTube viewership data, finding positive but varying correlations across artist groups.\"}]","Understanding Machine Taste - Aesthetic Agreement Across Foundation Models in K-pop | PDF",43]