[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-580996-105":59,"doc-detail-580996-en":131},{"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":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","research-matters-issue-40-the-ethics-of-deploying-large-language-models-in-high-stakes-automarking","Research Matters - Issue 40 - The ethics of deploying Large Language Models in high-stakes automarking","","Large Language Models (LLMs) such as Claude, Gemini, and GPT have gained attention for possible use in exam marking, where accurate interpretation of written student responses is essential. This article explores three fictitious high-stakes automarking scenarios and analyzes ethical issues supported by LLM capabilities as of June 2025: explainability, bias, and adversarial attacks. It compares automarkers with human examiners and argues that human graders offer higher trustworthiness in these settings.",{"@graph":69,"@context":123},[70,84,106],{"@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/research-matters-issue-40-the-ethics-of-deploying-large-language-models-in-high-stakes-automarking/580996/",{"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/research-matters-issue-40-the-ethics-of-deploying-large-language-models-in-high-stakes-automarking/580996.png","ImageObject",300,407,{"name":92,"@type":93},"Rowan","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-15","2026-10-15",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",11,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What ethical issues does the article examine for LLM-based automarking?","Question",{"text":113,"@type":114},"It examines explainability, bias, and adversarial attacks in the context of applying LLM-based tools to high-stakes exam marking.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does the article compare LLM-based automarkers with human examiners?",{"text":118,"@type":114},"It compares automarkers to human examiners using the context of well-functioning human marking in high-stakes settings, focusing on ethical dimensions informed by research.",{"name":120,"@type":111,"acceptedAnswer":121},"What conclusion does the article reach about trustworthiness?",{"text":122,"@type":114},"The article argues that human examiners have a higher potential for trustworthiness than LLM-based automarkers in high-stakes contexts.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},580996,1792172474,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":140,"language":141,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":67,"update_tm":145,"read_time":146,"preview_page_count":4},1099514067415,"https://ap-avatar.wpscdn.com/avatar/100002539d78ffe74a7?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779092875211072502","Research Matters / 40  \nA Cambridge University Press & Assessment publication  \nISSN: 17 55-6031  \nJournal homepage: [https://www.cambridgeassessment.org.uk/our-research/all](https://www.cambridgeassessment.org.uk/our-research/all)published-resources/research-matters/  \nThe ethics of deploying Large Language Models in high-stakes automarking  \nFrank Morley and Emma Walland  \nTo cite this article: Morley, F., & Walland, E. (2025). The ethics of deploying Large Language Models in high-stakes automarking. Research Matters: A Cambridge University Press &  \nAssessment publication, 40, 72–92. [https://doi.org/1 0.1 7863/CAM.1 21 937](https://doi.org/1 0.1 7863/CAM.1 21 937)  \nTo link this article: [https://www.cambridgeassessment.org.uk/Images/research](https://www.cambridgeassessment.org.uk/Images/research)matters-40-the-ethics-of-deploying-large-language-models-in-high-stakesautomarking.pdf  \nAbstract:  \nThe recent development of Large Language Models (LLMs) such as Claude, Gemini, and GPT has led to widespread attention on potential applications of these models. Marking exams is a domain which requires the ability to interpret and evaluate student responses (often consisting of written text), and the potential for artificial intelligence (AI) tools based on LLMs to contribute to the marking process has been noted and researched.  \nThis article imagines three scenarios where LLM-based automarking is applied to a high-stakes exam context, and examines three ethical issues identified based on current LLM technology as of June 2025 (Aloisi, 2023; Morley et al., 2025). These scenarios allow us to discuss: (1 ) explainability, whether automarking decisions can be explained; (2) bias, whether automarkers could make biased decisions disadvantaging certain demographic groups; and (3) adversarial attacks, how vulnerable automarkers might be to actors exploiting their vulnerabilities in order to produce a higher mark. Within the scenarios, we compare automarkers to human examiners in the context of well-functioning human marking in high-stakes settings (based on research regarding exam board marking in England). This allows us to explore ethical dimensions of automarkers based on what we know about automatic and human marking. In particular, this article argues that human examiners have a higher potential for trustworthiness over LLM-based automarkers.  \nCambridge University Press & Assessment is committed to making its documents accessible in accordance with the WCAG 2.1 Standard. We’re always looking to improve the accessibility of our documents. If you find any problems or you think we’re not meeting accessibility requirements, contact our team:  \nResearch Division, [ResearchDivision@cambridge.org](ResearchDivision@cambridge.org)  \nIf you need this document in a different format contact us, telling us your name, email address and requirements and we will respond within 1 5 working days.  \n© Cambridge University Press & Assessment 2025  \nFull Terms & Conditions of access and use can be found at  \nT&C: Terms and Conditions | Cambridge University Press & Assessment  \n© Cambridge University Press & Assessment 2025  \nThe ethics of deploying Large Language Models in high-stakes automarking  \nFrank Morley (Technology) and Emma Walland (Research Division)  \nIntroduction  \nAutomarking techniques which incorporate Large Language Models (LLMs) show potential in terms of increasing accuracy in comparison to previous automarking methods (see Morley et al., 2025). This has led to more interest in deploying LLMsas automarkers. However, the literature on LLM-based automarkers focuses on improving accuracy, while not as much emphasis is placed on ethical issues such as explainability, bias, and adversarial attacks (Aloisi, 2023; Morley et al., 2025). These considerations are critical in assessment contexts, particularly in high-stakes settings where students’ grade outcomes are being decided. Similar issues have also been examined in other reports. In pa","cbCaioaT37F4k3VT","https://ap.wps.com/l/cbCaioaT37F4k3VT","pdf",644406,22,"English","# Introduction\n## Explainability, Bias, and Adversarial Attacks\n## Comparison with Human Examiners\n## Accessibility and Contact Information","[{\"question\":\"What ethical issues does the article examine for LLM-based automarking?\",\"answer\":\"It examines explainability, bias, and adversarial attacks in the context of applying LLM-based tools to high-stakes exam marking.\"},{\"question\":\"How does the article compare LLM-based automarkers with human examiners?\",\"answer\":\"It compares automarkers to human examiners using the context of well-functioning human marking in high-stakes settings, focusing on ethical dimensions informed by research.\"},{\"question\":\"What conclusion does the article reach about trustworthiness?\",\"answer\":\"The article argues that human examiners have a higher potential for trustworthiness than LLM-based automarkers in high-stakes contexts.\"}]","Research Matters - Issue 40 - The ethics of deploying Large Language Models in high-stakes automarking | PDF",1792038654,55]