[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-1-en-105":3,"doc-seo-260541-105":53,"doc-detail-260541-en":126},{"code":4,"msg":5,"data":6},0,"success",[7,14,19,24,29,34,39,44,49],{"id":8,"doc_module":9,"doc_module_name":10,"category_name":11,"show_sort_weight":12,"slug":13},11,1,"Template","Presentations",90,"presentations",{"id":15,"doc_module":9,"doc_module_name":10,"category_name":16,"show_sort_weight":17,"slug":18},12,"Resumes",80,"resumes",{"id":20,"doc_module":9,"doc_module_name":10,"category_name":21,"show_sort_weight":22,"slug":23},14,"Invoices",70,"invoices",{"id":25,"doc_module":9,"doc_module_name":10,"category_name":26,"show_sort_weight":27,"slug":28},15,"Posters",60,"posters",{"id":30,"doc_module":9,"doc_module_name":10,"category_name":31,"show_sort_weight":32,"slug":33},16,"Social Media",50,"social-media",{"id":35,"doc_module":9,"doc_module_name":10,"category_name":36,"show_sort_weight":37,"slug":38},17,"Forms",40,"forms",{"id":40,"doc_module":9,"doc_module_name":10,"category_name":41,"show_sort_weight":42,"slug":43},18,"Letters",30,"letters",{"id":45,"doc_module":9,"doc_module_name":10,"category_name":46,"show_sort_weight":47,"slug":48},21,"Paper Templates",5,"papers-templates",{"id":50,"doc_module":9,"doc_module_name":10,"category_name":51,"show_sort_weight":4,"slug":52},158,"General","general-158",{"code":4,"msg":54,"data":55},"ok",{"site_id":56,"language":57,"slug":58,"title":59,"keywords":60,"description":61,"schema_data":62,"social_meta":119,"head_meta":121,"extra_data":123,"updated_unix":125},105,"en","blinded-by-context-unveiling-the-halo-effect-of-mllm-in-ai-hiring","Blinded by Context - Unveiling the Halo Effect of MLLM in AI Hiring","","This study investigates the halo effect in AI-driven hiring evaluations using Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs). Experiments with hypothetical job applications test how model judgments are shaped by non-job-related cues, including extracurricular activities and social-media-style images. Likert-scale responses across competency dimensions show significant halo effects, especially in image-based evaluations, while text-based assessments demonstrate greater resistance to bias. Results indicate multimodal supplementary information can substantially sway recruitment decisions, underscoring risks for AI hiring systems.",{"@graph":63,"@context":118},[64,80,101],{"@type":65,"itemListElement":66},"BreadcrumbList",[67,71,74,77],{"item":68,"name":69,"@type":70,"position":9},"https://docshare.wps.com","Home","ListItem",{"item":72,"name":10,"@type":70,"position":73},"https://docshare.wps.com/template/",2,{"item":75,"name":11,"@type":70,"position":76},"https://docshare.wps.com/template/presentations/",3,{"item":78,"name":59,"@type":70,"position":79},"https://docshare.wps.com/template/blinded-by-context-unveiling-the-halo-effect-of-mllm-in-ai-hiring/260541/",4,{"url":78,"name":59,"@type":81,"image":82,"author":87,"headline":59,"publisher":90,"fileFormat":93,"inLanguage":57,"description":61,"dateModified":94,"datePublished":95,"encodingFormat":93,"isAccessibleForFree":96,"interactionStatistic":97},"DigitalDocument",{"url":83,"@type":84,"width":85,"height":86},"https://docshare.wps.com/thumbnails/blinded-by-context-unveiling-the-halo-effect-of-mllm-in-ai-hiring/260541.png","ImageObject",442,249,{"name":88,"@type":89},"Jiven","Person",{"url":68,"name":91,"@type":92},"DocShare","Organization","application/pdf","2026-10-31","2026-09-13",true,{"@type":98,"interactionType":99,"userInteractionCount":8},"InteractionCounter",{"@type":100},"ViewAction",{"@type":102,"mainEntity":103},"FAQPage",[104,110,114],{"name":105,"@type":106,"acceptedAnswer":107},"What is the halo effect being studied in AI hiring evaluations?","Question",{"text":108,"@type":109},"The halo effect refers to how an overall impression of a person can influence judgments about unrelated attributes. In hiring, this can inflate perceived competence based on cues that are not directly job-related.","Answer",{"name":111,"@type":106,"acceptedAnswer":112},"How do the experiments test the impact of multimodal information?",{"text":113,"@type":109},"The study uses hypothetical job applications with structured hiring evaluations. Models are tested with resumes alone and with additional multimodal cues such as extracurricular text, social-media-style images, and video clips.",{"name":115,"@type":106,"acceptedAnswer":116},"What do the findings say about where bias appears most strongly?",{"text":117,"@type":109},"Image-based supplementary information induces stronger halo effects than text. Most models also show resilience to bias in text-based assessments, though halo effects are observed in video-simulated interview settings as well.","https://schema.org",{"og:url":78,"og:type":120,"og:title":59,"og:site_name":91,"og:description":61},"article",{"robots":122,"canonical":78},"index,follow",{"doc_id":124,"site_id":56},260541,1791726549,{"code":4,"msg":5,"data":127},{"doc_id":124,"user_id":128,"nickname":88,"user_avatar":129,"doc_module":9,"category_id":8,"category_name":11,"doc_title":59,"doc_description":61,"doc_content":130,"file_id":131,"file_url":132,"file_type":133,"file_size":134,"view_count":8,"is_deleted":4,"is_public":9,"is_downloadable":9,"audit_status":9,"page_count":135,"language":136,"language_code":57,"site_id":56,"html_lang":57,"table_of_contents":137,"faqs":138,"seo_title":139,"seo_description":61,"update_tm":140,"read_time":30,"preview_page_count":4},1099513958607,"https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399","Blinded by Context: Unveiling the Halo Effect of MLLM in AI Hiring  \nKyusik Kim* and Jeongwoo Ryu* and Hyeonseok Jeon and Bongwon Suh  \nDepartment of Intelligence and Information,  \nSeoul National University  \n{kyu823, jeongwoo, ikidson, [bongwon}@snu.ac.kr](bongwon}@snu.ac.kr)  \nAbstract  \nThis study investigates the halo effect in AIdriven hiring evaluations using Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) . Through experiments with hypothetical job applications, we examined how these models’ evaluations are influenced by non-job-related information, including extracurricular activities and social media images. By analyzing models’ responses to Likert-scale questions across different competency dimensions, we found that AI models exhibit significant halo effects, particularly in image-based evaluations, while text-based assessments showed more resistance to bias. The findings demonstrate that supplementary multimodal information can substantially influence AI hiring decisions, highlighting potential risksin AI-based recruitment systems.  \n1 Introduction  \nJob recruitment processes have traditionally been the domain of human decision-makers, who evaluate candidates based on resumes, interviews, and other relevant information. With recent advancements in artificial intelligence, a growing movement has emerged to leverage AI to enhance the efficiency and objectivity of these evaluations.  \nIn particular, Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) are increasingly being positioned as automated evaluators in tasks that demand structured judgment and decision-making. These models are now being deployed across a range of domains—including finance (Wang et al., 2024b ; Babaei and Giudici, 2024), law (Cheong et al., 2024 ; Guha et al., 2023), peer review (Jin et al., 2024a ; Kostic et al., 2024), and recruitment (Ganet al., 2024 ; Du et al., 2024)—where they assess  \n*These authors contributed equally to this work and should be considered co-first authors.  \ncomplex multimodal inputs and generate evaluative outputs that may influence consequential outcomes.  \nHowever, the implementation of AI in these evaluative settings, particularly in recruitment, must be approached with caution due to inherent biases. As LLMs and MLLMs take on greater responsibilities, concerns about bias in their decision-making have intensified. Research has identified various forms of bias, including gender bias (Chen et al., 2024b ; Kumar et al., 2024), racial bias (Kumar et al., 2024 ; Howard et al., 2024b), and bias based on physical attributes (Jiang et al., 2024 ; Sathe et al., 2024) .  \nIn the context of job hiring, one particularly significant cognitive bias is the halo effect—a phenomenon in which an overall impression of an individual influences judgments about unrelated attributes (Nisbett and Wilson, 1977 ; Thorndike, 1920) . This bias is well-documented in human decision-making, where positive impressions based on appearance, personality, or background often lead to inflated assessments of professional competence (Leuthesser et al., 1995 ; Cooper, 1981 ; Verhulst et al., 2010 ; Tsui and Barry, 1986) .  \nWhile the halo effect has been extensively studied in human evaluations, its impact on AI-driven assessments remains largely unexplored. Elangovan et al. (2024) suggests that LLMs, due to their holistic information processing, may overvalue certain attributes based on irrelevant cues. Some studies also indicate that LLMs may favor high-status authors or familiar individuals (Jin et al., 2024a ; Liu et al., 2025) . As MLLMs become more prevalent in hiring processes, it is essential to examine whether multimodal inputs induce similar biases in AI-based evaluations.  \nIn this work, we present the first comprehensive study of the halo effect in AI-driven recruitment (Figure 1) . We designed a structured hiring framework covering multiple job roles and resumes of varying quality. To assess the","cbCaifKuLbXanJfY","https://ap.wps.com/l/cbCaifKuLbXanJfY","pdf",5070680,47,"English","# Abstract\n# Introduction\n# Related Work","[{\"question\":\"What is the halo effect being studied in AI hiring evaluations?\",\"answer\":\"The halo effect refers to how an overall impression of a person can influence judgments about unrelated attributes. In hiring, this can inflate perceived competence based on cues that are not directly job-related.\"},{\"question\":\"How do the experiments test the impact of multimodal information?\",\"answer\":\"The study uses hypothetical job applications with structured hiring evaluations. Models are tested with resumes alone and with additional multimodal cues such as extracurricular text, social-media-style images, and video clips.\"},{\"question\":\"What do the findings say about where bias appears most strongly?\",\"answer\":\"Image-based supplementary information induces stronger halo effects than text. Most models also show resilience to bias in text-based assessments, though halo effects are observed in video-simulated interview settings as well.\"}]","Blinded by Context - Unveiling the Halo Effect of MLLM in AI Hiring | PDF",1789343402]