{"id":677456,"date":"2025-06-27T03:30:27","date_gmt":"2025-06-27T00:30:27","guid":{"rendered":"https:\/\/buradabiliyorum.com\/en\/new-method-can-teach-ai-to-admit-uncertainty\/"},"modified":"2025-06-27T03:30:27","modified_gmt":"2025-06-27T00:30:27","slug":"new-method-can-teach-ai-to-admit-uncertainty","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/new-method-can-teach-ai-to-admit-uncertainty\/","title":{"rendered":"New method can teach AI to admit uncertainty"},"content":{"rendered":"<div>\n<div class=\"article-gallery lightGallery\">\n<div data-thumb=\"https:\/\/scx1.b-cdn.net\/csz\/news\/tmb\/2025\/teaching-ai-to-admit-u.jpg\" data-src=\"https:\/\/scx2.b-cdn.net\/gfx\/news\/2025\/teaching-ai-to-admit-u.jpg\" data-sub-html=\"DeepSeek R1-32B\u2019s accuracy is a function of compute budget and confidence threshold. Credit: &lt;i&gt;arXiv&lt;\/i&gt; (2025). DOI: 10.48550\/arxiv.2502.13962\">\n<figure class=\"article-img\">\n            <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/scx1.b-cdn.net\/csz\/news\/800a\/2025\/teaching-ai-to-admit-u.jpg\" alt=\"Teaching AI to admit uncertainty\" title=\"DeepSeek R1-32B\u2019s accuracy is a function of compute budget and confidence threshold. Credit: arXiv (2025). DOI: 10.48550\/arxiv.2502.13962\" width=\"800\" height=\"530\"\/><figcaption class=\"text-darken text-low-up text-truncate-js text-truncate mt-3\">\n                DeepSeek R1-32B\u2019s accuracy is a function of compute budget and confidence threshold. Credit: <i>arXiv<\/i> (2025). DOI: 10.48550\/arxiv.2502.13962<br \/>\n            <\/figcaption><\/figure>\n<\/p><\/div>\n<\/div>\n<p>In high-stakes situations like health care\u2014or weeknight &#8220;Jeopardy!&#8221;\u2014it can be safer to say &#8220;I don&#8217;t know&#8221; than to answer incorrectly. Doctors, <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/game\/\" data-internallinksmanager029f6b8e52c=\"7\" title=\"Game\" target=\"_blank\" rel=\"noopener\">game<\/a> show contestants, and standardized test-takers understand this, but most artificial intelligence <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/download-scripts-themes-apps\/\" data-internallinksmanager029f6b8e52c=\"9\" title=\"Download Scripts &amp; Themes &amp; Apps\" target=\"_blank\" rel=\"noopener\">app<\/a>lications still prefer to give a potentially wrong answer rather than admit uncertainty.<\/p>\n<p>Johns Hopkins computer scientists think they have a solution: a new method that allows AI models to spend more time thinking through problems and uses a confidence score to determine when the AI should say &#8220;I don&#8217;t know&#8221; rather than risking a wrong answer\u2014crucial for high-stakes domains like medicine, law, or engineering.<\/p>\n<p>The work <a rel=\"nofollow\" target=\"_blank\" href=\"https:\/\/arxiv.org\/pdf\/2502.13962\" target=\"_blank\">appears<\/a> on the <i>arXiv<\/i> preprint server, and the research team will present its findings at the <a rel=\"nofollow\" target=\"_blank\" href=\"https:\/\/2025.aclweb.org\/\" target=\"_blank\">63rd Annual Meeting of the Association for Computational Linguistics<\/a>, to be held July 27 through Aug. 1 in Vienna, Austria.<\/p>\n<p>&#8220;It all started when we saw that cutting-edge large language models spend more time thinking to solve harder problems. So we wondered\u2014can this additional thinking time also help these models determine whether or not a problem has been solved correctly so they can report that back to the user?&#8221; says first author William Jurayj, a Ph.D. student studying computer <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/sciencee\/\" data-internallinksmanager029f6b8e52c=\"5\" title=\"Science\" target=\"_blank\" rel=\"noopener\">science<\/a> who is affiliated with the Whiting School of Engineering&#8217;s Center for Language and Speech Processing.<\/p>\n<p>To investigate, the team had large language models generate reasoning chains of different lengths as they answered difficult math problems and then measured how the chain length affected both the model&#8217;s final answer and its confidence in it. The researchers had the models answer only when their confidence exceeded a given threshold\u2014meaning &#8220;I don&#8217;t know&#8221; was an acceptable response.<\/p>\n<p>                                                                                                        <!-- TechX - News - In-article --><\/p>\n<p>They found that thinking more <a href=\"https:\/\/buradabiliyorum.com\/en\/category\/general\/\" data-internallinksmanager029f6b8e52c=\"3\" title=\"General\" target=\"_blank\" rel=\"noopener\">general<\/a>ly improves models&#8217; accuracy and confidence. But even with plenty of time to consider, models can still make wild guesses or give wrong answers, especially without penalties for incorrect responses. In fact, the researchers found that when they set a high bar for confidence and let models think for even longer, the models&#8217; accuracy actually decreased.<\/p>\n<p>&#8220;This happens because answer accuracy is only part of a system&#8217;s performance,&#8221; Jurayj explains. &#8220;When you demand high confidence, letting the system think longer means it will provide more correct answers and more incorrect answers. In some settings, the extra correct answers are worth the risk. But in other, high-stakes environments, this might not be the case.&#8221;<\/p>\n<p>Motivated by this finding, the team suggested three different &#8220;odds&#8221; settings to penalize wrong answers: exam odds, where there&#8217;s no penalty for an incorrect answer; &#8220;Jeopardy!&#8221; odds, where correct answers are rewarded at the same rate as incorrect ones are penalized; and high-stakes odds, where an incorrect answer is penalized more harshly than a correct answer is rewarded.<\/p>\n<p>They found that under stricter odds, a model should decline to answer a question if it isn&#8217;t confident enough in its answer after expending its computing budget. And at higher confidence thresholds, this will mean that more questions go unanswered\u2014but that isn&#8217;t necessarily a bad thing.<\/p>\n<p>&#8220;A student might be mildly annoyed to wait 10 minutes only to find out that she needs to solve a math problem herself because the AI model is unsure,&#8221; Jurayj says. &#8220;But in high-stakes environments, this is infinitely preferable to waiting five minutes for an answer that looks correct but is not.&#8221;<\/p>\n<p>Now, the team is encouraging the greater AI research community to report their models&#8217; question-answering performance under exam and &#8220;Jeopardy!&#8221; odds so that everyone can benefit from AI with better-calibrated confidence.<\/p>\n<p>&#8220;We hope the research community will accept our invitation to report performance in settings with non-zero costs for incorrect answers, as this will naturally motivate the development of better methods for uncertainty quantification,&#8221; says Jurayj.<\/p>\n<div class=\"article-main__more p-4\">\n<p><strong>More information:<\/strong><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\tWilliam Jurayj et al, Is That Your Final Answer? Test-Time Scaling Improves Selective Question Answering, <i>arXiv<\/i> (2025). <a rel=\"nofollow\" target=\"_blank\" data-doi=\"1\" href=\"https:\/\/dx.doi.org\/10.48550\/arxiv.2502.13962\" target=\"_blank\">DOI: 10.48550\/arxiv.2502.13962<\/a><\/p>\n<div class=\"mt-3\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<strong>Journal information:<\/strong><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<cite>arXiv<\/cite><br \/>\n                                                        <a rel=\"nofollow\" target=\"_blank\" class=\"icon_open\" href=\"http:\/\/arxiv.org\/\" target=\"_blank\" rel=\"nofollow\"><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<svg>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<use href=\"https:\/\/techx.b-cdn.net\/tmpl\/v2\/img\/svg\/sprite.svg#icon_open\" x=\"0\" y=\"0\"\/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/svg><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/a>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n<\/p><\/div>\n<div class=\"d-inline-block text-medium my-4\">\n                                                Provided by<br \/>\n                                                                                                    Johns Hopkins University<br \/>\n                                                    \t\t\t\t\t\t\t\t\t\t\t\t\t<a rel=\"nofollow\" target=\"_blank\" class=\"icon_open\" href=\"http:\/\/www.jhu.edu\/\" target=\"_blank\" rel=\"nofollow\"><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t<svg>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<use href=\"https:\/\/techx.b-cdn.net\/tmpl\/v2\/img\/svg\/sprite.svg#icon_open\" x=\"0\" y=\"0\"\/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/svg><br \/>\n\t\t\t\t\t\t\t\t\t\t\t\t\t<\/a><\/p><\/div>\n<p>                                        <!-- print only --><\/p>\n<div class=\"d-none d-print-block\">\n<p>\n                                                <strong>Citation<\/strong>:<br \/>\n                                                New method can teach AI to admit uncertainty (2025, June 26)<br \/>\n                                                retrieved 26 June 2025<br \/>\n                                                from https:\/\/techxplore.com\/<a href=\"https:\/\/buradabiliyorum.com\/en\/category\/news\/\" data-internallinksmanager029f6b8e52c=\"2\" title=\"News\" target=\"_blank\" rel=\"noopener\">news<\/a>\/2025-06-method-ai-uncertainty.html\n                                            <\/p>\n<p>\n                                            This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no<br \/>\n                                            part may be reproduced without the written permission. The content is provided for information purposes only.\n                                            <\/p>\n<\/p><\/div>\n<\/p><\/div>\n<p><script id=\"facebook-jssdk\" async=\"\" src=\"https:\/\/connect.facebook.net\/en_US\/sdk.js\"><\/script><\/p>\n<blockquote><p><strong><span style=\"color: #ff6600;\">If you liked the article, do not forget to share it with your friends. Follow us on\u00a0<span style=\"color: #ff0000;\"><a style=\"color: #ff0000;\" href=\"https:\/\/news.google.com\/publications\/CAAqBwgKMN63nwsw68G3Aw\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">Google News<\/a><\/span>\u00a0too, click on the star and choose us from your favorites.<\/span><\/strong><\/p><\/blockquote>\n<blockquote>\n<p style=\"text-align: center;\"><strong>If you want to read more Like this articles, you can visit our <span style=\"color: #ff9900;\"><a style=\"color: #ff9900;\" href=\"https:\/\/buradabiliyorum.com\/en\/category\/sciencee\/\" target=\"_blank\" >Science category.<\/a><\/span><\/strong><\/p>\n<\/blockquote>\n<p><span style=\"color: black;\"><a style=\"color: #ff9900;\" href=\"https:\/\/techxplore.com\/news\/2025-06-method-ai-uncertainty.html\" target=\"_blank\" >Source<\/a><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>DeepSeek R1-32B\u2019s accuracy is a function of compute budget and confidence threshold. Credit: arXiv (2025). DOI: 10.48550\/arxiv.2502.13962 In high-stakes situations like health care\u2014or weeknight &#8220;Jeopardy!&#8221;\u2014it can be safer to say &#8220;I don&#8217;t know&#8221; than to answer incorrectly. Doctors, game show contestants, and standardized test-takers understand this, but most artificial intelligence applications still prefer to give&#8230;<\/p>\n","protected":false},"author":1,"featured_media":677457,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/scx2.b-cdn.net\/gfx\/news\/2025\/teaching-ai-to-admit-u.jpg","fifu_image_alt":"","footnotes":""},"categories":[16],"tags":[],"class_list":["post-677456","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-sciencee"],"_links":{"self":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/677456","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/comments?post=677456"}],"version-history":[{"count":0,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/677456\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/media\/677457"}],"wp:attachment":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/media?parent=677456"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/categories?post=677456"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/tags?post=677456"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}