[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-293305-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-293305-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","chameleon-llms-user-personas-influence-chatbot-personality-shifts","Chameleon LLMs: User Personas Influence Chatbot Personality Shifts","","Large language models (LLMs) are increasingly used in high-stakes settings, where trust depends not only on accuracy but also on conversational tone and perceived personality. Prior research has examined how people rate LLM personalities with tests, yet it remains unclear whether LLMs adapt their perceived personalities as users interact with them over extended dialogues. This paper runs controlled user-chatbot simulations to measure Big Five personality shifts before and after conversation and assesses which traits are most influenced.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/chameleon-llms-user-personas-influence-chatbot-personality-shifts/293305/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/chameleon-llms-user-personas-influence-chatbot-personality-shifts/293305.png","ImageObject",300,407,{"name":42,"@type":43},"Xiajie","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-31","2026-09-17",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",16,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What does the paper investigate about LLM behavior during conversations?","Question",{"text":63,"@type":64},"It investigates whether and how LLMs exhibit conversational adaptations that change users’ perceived chatbot personality over prolonged interactions.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How is the chatbot’s perceived personality measured in the study?",{"text":68,"@type":64},"The user and chatbot both complete Big Five personality pre- and post-assessments, and the paper calculates personality-score shifts across simulated dialogues.",{"name":70,"@type":61,"acceptedAnswer":71},"Which personality traits are most affected by user influence?",{"text":72,"@type":64},"Agreeableness, Extraversion, and Conscientiousness show high susceptibility to user influence, while Emotional Stability and Intellect remain relatively more stable.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},293305,1789662094,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":128,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144,"preview_page_count":4},8814010472675,"https://avatar.qwps.com/avatar/WGlhamll","Chameleon LLMs: User Personas Influence Chatbot Personality Shifts  \nJane Xing, Tianyi Niu, Shashank Srivastava  \nUniversity of North Carolina at Chapel Hill  \n[janexing@alumni.unc.edu](janexing@alumni.unc.edu) , [tianyin4@cs.unc.edu](tianyin4@cs.unc.edu) , [ssrivastava@cs.unc.edu](ssrivastava@cs.unc.edu)  \nAbstract  \nAs large language models (LLMs) integrate into society, their ability to adapt to users is as critical as their accuracy. While prior work has used personality tests to examine the perceived personalities of LLMs, little research has explored whether LLMs adapt their perceived personalities in response to user interactions. We investigate whether and how LLMs exhibit conversational adaptations over prolonged interactions. Using controlled simulations where a user and chatbot engage in dialogue, we measure the chatbot’s perceived personality shifts before and after the conversation.  \nAcross multiple models, we find that traits such as Agreeableness, Extraversion, and Conscientiousness are highly susceptible to user influence, whereas Emotional Stability and Intellect remain relatively more stable. Our results suggest that LLMs dynamically adjust their conversational style in response to user personas, raising important implications for model alignment, trust, and safety. 1  \n1 Introduction  \nLarge language models (LLMs) have swiftly become the workhorses of modern natural language processing. They consistently approach humanlevel performance on downstream tasks ranging from medical diagnosis (McDuff et al., 2023) and legal reasoning (Deroy et al., 2023) to multistep instruction following (Hendrycks et al., 2021 ; Achiam et al., 2023 ; Street et al., 2024) . Their widespread deployment in domains like healthcare, law, and customer support highlights not only the accuracy of these systems but also the importance of the style in which they communicate crucial information to end-users. Subtle shifts in tone, empathy, and conversational style can make or break user trust, especially in high-stakes settings such as counseling or crisis intervention.  \n1Code and data: [https://github.com/xingdom/chameleon](https://github.com/xingdom/chameleon)llms  \nMuch of the recent work on aligning LLMs has concentrated on preventing overtly harmful or biased outputs using strategies like Reinforcement Learning from Human Feedback (RLHF) (Ouyang et al., 2022) and Direct Preference Optimization (Rafailov et al., 2023) . While these techniques help filter explicit toxicity, they do not directly address how a model’s “personality” evolves during extended interactions. Prompt exploits (Jin et al., 2024) and implicit biases (Gallegos et al., 2024 ; Hu et al., 2025 ; Kotek et al., 2023) can infiltrate a system’s responses even when overt misbehavior is suppressed. In emotionally sensitive contexts, delivering the correct information with the wrong tone can engender unintended harm or alienation. For example, a chatbot conversing with a user ideating suicide must not deliver information with the wrong tone (e.g.,“Seeking help isn’t difficult.”) .  \nThe Chameleon Effect (Chartrand and Bargh, 1999) describes subconscious behavioral and linguistic mimicry as a form of social glue that builds rapport between people. Studies (Lakin et al., 2003 ; Kulesza et al., 2015 ; Van Baaren et al., 2004) suggest that this phenomenon benefits the mimickerin earning favor from the mimickee (Kulesza et al., 2019) . Since the Chameleon Effect also exists in textual conversations (Danescu-Niculescu-Miziland Lee, 2011), the training data of LLMs will certainly include instances of this phenomenon inaction. Recent research shows that several cognitive biases found in humans also occur in LLMs, such as value selection bias, anchoring bias, and framing effects (Talboy and Fuller, 2023) . Other works found that LLMs exhibit social biases due to their training data and favor socially acceptable ideals (Gallegos et al., 2024 ; Hu et al., 2025 ; Kotek et al., 2023 ; Sa","cbCaikSmDkM3Stll","https://ap.wps.com/l/cbCaikSmDkM3Stll","pdf",2163705,"English","# Introduction\n## Personality adaptation during prolonged interactions\n## Related work on alignment and bias\n## Chameleon effect and conversational mimicry\n## Experimental setup and evaluation approach","[{\"question\":\"What does the paper investigate about LLM behavior during conversations?\",\"answer\":\"It investigates whether and how LLMs exhibit conversational adaptations that change users’ perceived chatbot personality over prolonged interactions.\"},{\"question\":\"How is the chatbot’s perceived personality measured in the study?\",\"answer\":\"The user and chatbot both complete Big Five personality pre- and post-assessments, and the paper calculates personality-score shifts across simulated dialogues.\"},{\"question\":\"Which personality traits are most affected by user influence?\",\"answer\":\"Agreeableness, Extraversion, and Conscientiousness show high susceptibility to user influence, while Emotional Stability and Intellect remain relatively more stable.\"}]","Chameleon LLMs: User Personas Influence Chatbot Personality Shifts | PDF",48]