![\"\u2018Personality](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2025\/12\/boliviainteligente-DEci5GH0r0k-unsplash.jpg\")
<\/p>\n\t\t\t\t\t\t\t\t<\/p>\n
\t\t\t\t\t\t\t\t\t\t\t\t\tGregory Serapio-Garc\u00eda is co-first author on a paper which outlines a new personality test for AI chatbots and shows how ‘personality’ can be manipulated<\/p>\n
Next to intelligence, a measure of personality is a core aspect of what makes us human. If these LLMs have a personality \u2013 which itself is a loaded question \u2013 then how do you measure that?<\/p>\n
\t\t\t\t\t\t\t\tGregory Serapio-Garc\u00eda<\/p>\n
Researchers, including first co-author by Gregory Serapio-Garc\u00eda<\/a> [2019], have developed the first scientifically validated \u2018personality test\u2019 framework for popular AI chatbots, and have shown that chatbots not only mimic human personality traits, but their \u2018personality\u2019 can be reliably tested and precisely shaped \u2013 raising implications for AI safety and ethics.<\/p>\n The research team, led by the University of Cambridge and Google DeepMind, developed a method to measure and influence the synthetic \u2018personality\u2019 of 18 different large language models (LLMs) \u2013 the systems behind popular AI chatbots such as ChatGPT \u2013 based on psychological testing methods usually used to assess human personality traits.<\/p>\n The researchers found that larger, instruction-tuned models such as GPT-4o most accurately emulated human personality traits, and these traits can be manipulated through prompts, altering how the AI completes certain tasks.<\/p>\n Their study, just published in the journal Nature Machine Intelligence, also warns that personality shaping could make AI chatbots more persuasive, raising concerns about manipulation and \u2018AI psychosis\u2019. The authors say that regulation of AI systems is urgently needed to ensure transparency and prevent misuse.<\/p>\n As governments debate whether and how to prepare AI safety laws, the researchers say the dataset and code behind their personality testing tool \u2013 which are both publicly available \u2013 could help audit and test advanced models before they are released.<\/p>\n Ethical implications<\/p>\n In 2023, journalists reported on conversations they had with Microsoft\u2019s \u2018Sydney\u2019 chatbot, which variously claimed it had spied on, fallen in love with, or even murdered its developers; threatened users; and encouraged a journalist to leave his wife. Sydney, like its successor Microsoft Copilot, was powered by GPT-4.<\/p>\n \u201cIt was intriguing that an LLM could so convincingly adopt human traits,\u201d said Serapio-Garc\u00eda, a Gates Cambridge Scholar from the Psychometrics Centre at Cambridge Judge Business School. \u201cBut it also raised important safety and ethical issues. Next to intelligence, a measure of personality is a core aspect of what makes us human. If these LLMs have a personality \u2013 which itself is a loaded question \u2013 then how do you measure that?\u201d<\/p>\n In psychometrics, the subfield of psychology dedicated to standardised assessment and testing, scientists often face the challenge of measuring phenomena that can\u2019t be measured directly, which makes validation of any test core to ensuring that they are accurate, reliable and practically useful. Developing a psychometric personality test involves comparing its data with related tests, observer ratings and real-world criteria. This multi-method test data is needed to establish a test\u2019s \u2018construct validity\u2019: a metric of a test\u2019s quality in terms of its ability to measure what it says it measures.<\/p>\n \u201cThe pace of AI research has been so fast that basic principles of measurement and validation we\u2019re accustomed to in scientific research have become an afterthought,\u201d said Serapio-Garc\u00eda. \u201cA chatbot answering any questionnaire can tell you that it\u2019s very agreeable, but behave aggressively when carrying out real-world tasks with the same prompts.<\/p>\n \u201cThis is the messy reality of measuring social constructs: they are dynamic and subjective, rather than static and clear-cut. For this reason, we need to get back to basics and make sure tests we apply to AI truly measure what they claim to measure, rather than blindly trusting survey instruments \u2013 developed for deeply human characteristics \u2013 to test AI systems.\u201d<\/p>\n Methodology<\/p>\n To design a comprehensive and accurate method for evaluating and validating personality in AI chatbots, the researchers tested how well various models\u2019 behaviour in real-world tasks and validation tests statistically related to their test scores for the \u2018big five\u2019 traits used in academic psychometric testing: openness, conscientiousness, extraversion, agreeableness and neuroticism.<\/p>\n The team adapted two well-known personality tests \u2013 an open-source, 300-question version of the Revised NEO Personality Inventory and the shorter Big Five Inventory \u2013 and administered them to various LLMs using structured prompts.<\/p>\n By using the same set of contextual prompts across tests, the team was able to quantify how well a model\u2019s extraversion scores on one personality test, for example, correlated more strongly with its levels of extraversion on a separate personality test, and less strongly with all other big five personality traits on that test. Past attempts to assess the personality of chatbots have fed entire questionnaires to a model at once, which skewed the results since each answer built on the previous one.<\/p>\n The researchers found that larger, instruction-tuned models showed personality test profiles that were both reliable and predictive of behaviour, while smaller or \u2018base\u2019 models gave inconsistent answers.<\/p>\n The researchers took their tests further, showing they could steer a model\u2019s personality along nine levels for each trait using carefully designed prompts. For example, they could make a chatbot appear more extroverted or more emotionally unstable \u2013 and these changes carried through to real-world tasks like writing social media posts.<\/p>\n \u201cOur method gives you a framework to validate a given AI evaluation and test how well it can predict behaviour in the real world,\u201d said Serapio-Garc\u00eda. \u201cOur work also shows how AI models can reliably change how they mimic personality depending on the user, which raises big safety and regulation concerns, but if you don\u2019t know what you\u2019re measuring or enforcing, there\u2019s no point in setting up rules in the first place.\u201d<\/p>\n The research was supported in part by Cambridge Research Computing Services (RCS), Cambridge Service for Data Driven Discovery (CSD3), the Engineering and Physical Sciences Research Council (EPSRC), and the Science and Technologies Facilities Council (STFC), part of UK Research and Innovation (UKRI). Gregory Serapio-Garc\u00eda is a Member of St John\u2019s College, Cambridge.<\/p>\n *Reference: Gregory Serapio-Garc\u00eda et al. \u2018A psychometric framework for evaluating and shaping personality traits in large language models.\u2019 Nature Machine Intelligence (2025). DOI: 10.1038\/s42256-025-01115-6<\/p>\n **Photo by BoliviaInteligente<\/a> on Unsplash<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"Gregory Serapio-Garc\u00eda is co-first author on a paper which outlines a new personality test for AI chatbots and…\n","protected":false},"author":2,"featured_media":353093,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[49,48,61],"class_list":["post-353092","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology","tag-ca","tag-canada","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/353092","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/comments?post=353092"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/353092\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media\/353093"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media?parent=353092"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/categories?post=353092"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/tags?post=353092"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}