[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-620928-105":3,"detail-sidebar-cat-0-en-105":80,"doc-detail-620928-en":130},{"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":73,"head_meta":75,"extra_data":77,"updated_unix":79},105,"en","learning-to-live-with-ai-in-high-stakes-assessment","Learning to live with AI in high-stakes assessment","","Learning to live with AI in high-stakes assessment explores how artificial intelligence could transform marking, re-mark appeals, and item writing within exam contexts. The research emphasizes key obstacles: limited explainability, unreliable judgments that cannot separate relevant content or fact from fiction, and ethical risks from embedded societal bias requiring continuous monitoring. It outlines prerequisites for integration, including public trust, regulation, accountability and expertise embedding.",{"@graph":14,"@context":72},[15,34,55],{"@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/exam/","Exam",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/learning-to-live-with-ai-in-high-stakes-assessment/620928/",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/learning-to-live-with-ai-in-high-stakes-assessment/620928.png","ImageObject",300,407,{"name":42,"@type":43},"Jordan Avery","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-27","2026-10-20",true,{"@type":52,"interactionType":53,"userInteractionCount":22},"InteractionCounter",{"@type":54},"ViewAction",{"@type":56,"mainEntity":57},"FAQPage",[58,64,68],{"name":59,"@type":60,"acceptedAnswer":61},"How can AI support marking and re-mark processes in high-stakes assessment?","Question",{"text":62,"@type":63},"AI has potential to mark some types of exam questions and to support the re-mark process when a student appeals the grade awarded. It may also assist with item writing.","Answer",{"name":65,"@type":60,"acceptedAnswer":66},"Why is explainability a critical challenge for AI in assessment?",{"text":67,"@type":63},"An AI system cannot explain or justify its judgments in an acceptable way, lacking real-world context beyond training data. It may highlight used keywords, but it cannot evaluate argument quality like a human marker.",{"name":69,"@type":60,"acceptedAnswer":70},"What ethical risks arise when training AI systems for assessment?",{"text":71,"@type":63},"AI systems learn and reflect wider societal attitudes present in their training datasets. This can lead to favoritism or discrimination based on characteristics such as sex, ethnicity, or religious beliefs, requiring ongoing monitoring and correction.","https://schema.org",{"og:url":32,"og:type":74,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":76,"canonical":32},"index,follow",{"doc_id":78,"site_id":7},620928,1793136613,{"code":4,"msg":81,"data":82},"success",[83,87,91,94,99,104,109,114,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":84,"show_sort_weight":85,"slug":86},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":88,"show_sort_weight":89,"slug":90},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":92,"slug":93},70,"exam",{"id":95,"doc_module":4,"doc_module_name":25,"category_name":96,"show_sort_weight":97,"slug":98},5,"Comic",60,"comic",{"id":100,"doc_module":4,"doc_module_name":25,"category_name":101,"show_sort_weight":102,"slug":103},6,"Technology",50,"technology",{"id":105,"doc_module":4,"doc_module_name":25,"category_name":106,"show_sort_weight":107,"slug":108},7,"Healthcare",40,"healthcare",{"id":110,"doc_module":4,"doc_module_name":25,"category_name":111,"show_sort_weight":112,"slug":113},8,"Research & Report",30,"research-report",{"id":115,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},9,"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":95,"slug":129},19,"General","general",{"code":4,"msg":81,"data":131},{"doc_id":78,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":33,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":22,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":26,"language":139,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":12,"update_tm":143,"read_time":95,"preview_page_count":4},1099523882367,"https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc","Learning to live with AI in high-stakes assessment  \nArtificial intelligence (AI) technology is set to revolutionise how we work and live. This includes its potential to drive creative change in the educational sector. However, AI is accelerating much faster than our capacity to understand and regulate it .  \nOur research focus  \nAt AQA, we’ve been researching the use of AI in assessment for some time. We know that AI has the potential to mark some types of exam questions and to support the re-mark process, where a student has appealed the grade awarded to them. We’ve also considered how AI might assist with item writing.  \nOur work has highlighted a number of challenges to overcome if we want to steer AI for positive change.  \nChallenges to using AI systems They can’t explain their decisions  \nAn AI system can’t explain or justify the basis for its judgements in a way that we would find acceptable. They have no real-world context to draw upon, beyond the data they have been trained on. For example, an AI system can highlight the keywords in a student’s response that it has used to determine the mark, but it can’t evaluate the quality of an argument as a human marker could.  \nThey are unreliable  \nStandard AI technologies are unable to tell the difference between important or unimportant content in a text. Newer generative AI technologies can’t tell fact from fiction, and often repeat falsehoods.  \nTraining them to behave ethically is not easy  \nAI systems ‘learn’ and reflect back wider societal attitudes that already exist in the data sets they’re trained on. This means an AI system could treat some groups of people more favourably or discriminate against them, based on characteristics such as sex, ethnicity or religious beliefs. Continuous monitoring and correcting would be needed to prevent this.  \nWhat will it take to integrate AI into high-stakes assessment?  \nPublic trust  \nTrustworthy AI  Trusted people   \nTrust-reinforcing regulations   \nEthical AI use in assessment  \nQuestions around trust and accountability  \n• Who is responsible for the AI’s ‘behaviour’?  \n• Who should be expected to quality-assure the AI system?  \n• Who should have the final say in the case of a marking review – the human marker or the AI?  \n• Who holds the intellectual property of AI-generated content?  \nEmbedding assessment expertise  \nThe AI developers who train, assess and evaluate their systems are not experts in assessment. This skills gap can lead to unrealistic claims of AI competencies. Given that these systems will be used to perform tasks alongside or in place of people, it is vital to embed the right expertise at an early stage.  \nLooking to the future  \nAI holds great promise for an exciting future, with the potential to develop more personalised educational tools as well as opportunities to enhance writing and marking processes. However, research will need to focus on the issues of unreliability, bias and low explainability to maintain transparency and accountability in the exam system.  \nCopyright © 2023 AQA and its licensors. All rights reserved. G02516-02c","cbCaioATbpQXfkUa","https://ap.wps.com/l/cbCaioATbpQXfkUa","pdf",250432,"English","# Our research focus\n# Challenges to using AI systems\n## Explainability limits\n## Reliability concerns\n## Ethical training and bias\n# What will it take to integrate AI into high-stakes assessment?\n## Public trust and regulations\n## Accountability questions\n## Embedding assessment expertise\n# Looking to the future","[{\"question\":\"How can AI support marking and re-mark processes in high-stakes assessment?\",\"answer\":\"AI has potential to mark some types of exam questions and to support the re-mark process when a student appeals the grade awarded. It may also assist with item writing.\"},{\"question\":\"Why is explainability a critical challenge for AI in assessment?\",\"answer\":\"An AI system cannot explain or justify its judgments in an acceptable way, lacking real-world context beyond training data. It may highlight used keywords, but it cannot evaluate argument quality like a human marker.\"},{\"question\":\"What ethical risks arise when training AI systems for assessment?\",\"answer\":\"AI systems learn and reflect wider societal attitudes present in their training datasets. This can lead to favoritism or discrimination based on characteristics such as sex, ethnicity, or religious beliefs, requiring ongoing monitoring and correction.\"}]","Learning to live with AI in high-stakes assessment | PDF",1792464760]