{"id":61317,"date":"2024-05-29T10:56:46","date_gmt":"2024-05-29T10:56:46","guid":{"rendered":"https:\/\/news.talkwithrattan.com\/index.php\/2024\/05\/29\/a-new-test-could-help-weed-out-ai-generated-text\/"},"modified":"2024-05-29T10:56:46","modified_gmt":"2024-05-29T10:56:46","slug":"a-new-test-could-help-weed-out-ai-generated-text","status":"publish","type":"post","link":"https:\/\/news.talkwithrattan.com\/index.php\/2024\/05\/29\/a-new-test-could-help-weed-out-ai-generated-text\/","title":{"rendered":"A new test could help weed out AI-generated text"},"content":{"rendered":"<div style=\"text-align:center\"><img loading=\"lazy\" decoding=\"async\" width=\"592\" height=\"240\" src=\"https:\/\/i2.wp.com\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module-sm@2x.png?resize=592,240&amp;ssl=1\" class=\"attachment-post-thumbnail size-post-thumbnail wp-post-image\" alt=\"A new test could help weed out AI-generated text\" title=\"A new test could help weed out AI-generated text\" \/><\/div><p> <br \/>\n<\/p>\n<div data-component=\"video-embed\">\n<p>Imagine that you\u2019re helping judge a writing contest at your school. You want to make sure everyone did their own work. If someone used an <a href=\"https:\/\/www.snexplores.org\/article\/scientists-say-artificial-intelligence-definition-pronunciation\">artificial intelligence<\/a>, or AI, model such as ChatGPT to write an entry, that shouldn\u2019t count. But how can you tell whether something was written by AI? New research reveals <a href=\"https:\/\/arxiv.org\/pdf\/2401.12970.pdf\" rel=\"noopener\">a simple way to test whether a person wrote something or not<\/a>. Just ask a bot to rewrite it.<\/p>\n<p>\u201cIf you ask AI to rewrite content written by AI, it will have very few edits,\u201d says Chengzhi Mao. When AI rewrites a person\u2019s text, it typically makes many more changes.<\/p>\n<aside class=\"sn-conversion rich-text alignright\"\/>\n<p>Mao, Junfeng Yang and their colleagues designed a tool called Raidar. It\u2019s a detector that uses AI rewriting to detect bot-generated text. Mao is a researcher at the Software Systems Lab at Columbia University in New York. Yang leads this lab.<\/p>\n<p>Separating bot-talk from person-talk is \u201cvery important,\u201d says Yang. Lots of AI writing has already flooded social media and product reviews. It has fueled fake news websites and spam books. Some <a href=\"https:\/\/www.snexplores.org\/article\/chatgpt-homework-school-help-learning-ai-bot-mistakes\">students use AI to cheat on homework and tests<\/a>. Tools like Raidar could help expose AI-powered cheaters and liars.<\/p>\n<p>Raidar\u2019s creators shared the tool at the International Conference on Learning Representations. That meeting was on May 7 in Vienna, Austria.<\/p>\n<section class=\"newsletter-signup__wrapper___lZ0W1 wp-block-house-ads wp-block-newsletter-signup\">\n<picture><source srcset=\"https:\/\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module@1x.png 1x,&#10;&#9;&#9;&#9;&#9;https:\/\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module@2x.png 2x\" media=\"(min-width: 768px)\"><source srcset=\"https:\/\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module-sm@1x.png 1x,&#10;&#9;&#9;&#9;&#9;https:\/\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module-sm@2x.png 2x\"><\/source><\/source><\/picture>\n<div class=\"newsletter-signup__container___srNOL\" data-component=\"newsletter-signup\">\n<h3 class=\"newsletter-signup__heading___0EHmb\">\n\t\t\tEducators and Parents, Sign Up for The Cheat Sheet\t\t<\/h3>\n<div class=\"newsletter-signup__message___pemaq\">\n<p>Weekly updates to help you use <em>Science News Explores<\/em> in the learning environment<\/p>\n<\/p><\/div>\n<p class=\"newsletter-signup__thankyou___K6GGN\">Thank you for signing up!<\/p>\n<p class=\"newsletter-signup__error___hCsJI\">There was a problem signing you up.<\/p>\n<\/p><\/div>\n<\/section>\n<h4 class=\"wp-block-heading\">Weeding out AI<\/h4>\n<p>Mao regularly uses ChatGPT to help polish his own writing. For example, he sometimes asks the bot to rewrite and improve an email. He noticed that this bot can do a pretty good job the first time it rewrites something that he wrote. But if he asks it to improve an email again \u2014 revising its own bot writing \u2014 then it won\u2019t change much.<\/p>\n<p>\u201cThat\u2019s how we got motivated,\u201d Mao says. He realized the number of edits a bot makes to a piece of writing might say something about how the original text got written.<\/p>\n<p>\u201cIt\u2019s a pretty neat idea,\u201d says Amrita Bhattacharjee. \u201cNobody had thought of it before.\u201d Bhattacharjee is a PhD student at Arizona State University in Tempe. She has researched <a href=\"https:\/\/arxiv.org\/abs\/2308.01284\" rel=\"noopener\">AI-generated text detection<\/a>, but wasn\u2019t involved in developing Raidar.<\/p>\n<aside class=\"sn-conversion rich-text\"\/>\n<p>Raidar is a tool that determines whether text was likely AI-generated or not. To test Raidar, the team gathered writing samples from people and several different chatbots. They did this for a few kinds of text, including news, Yelp reviews, student essays and computer <a href=\"https:\/\/www.snexplores.org\/article\/scientists-say-code-definition-pronunciation\">code<\/a>. Then, the team had several AI models rewrite all the human-written and bot-written samples.<\/p>\n<p>Next, the researchers used a simple computer program to calculate the number of changes between the original and edited version of each writing sample. This step doesn\u2019t require any AI. Based on the number of changes in a revision, Raidar could sort writing samples into human-generated and AI-generated. This worked well even if the AI that did the rewriting was different from the AI that wrote the original sample.<\/p>\n<p>Raidar\u2019s sorting is not perfect. The tool sometimes identifies a human text as AI, or vice versa. But it performs better than other tools designed to detect AI-written text, the researchers found.<\/p>\n<p>Most other tools use AI models and statistics to learn to recognize the kind of text that bots produce. These tools typically work best on longer passages of text. They may not work at all on short blurbs, like the ones found on social media or homework assignments. But Raidar works well even on text that\u2019s just 10 words long.<\/p>\n<figure class=\"wp-block-image alignwide size-full\" data-component=\"lightbox\"><a href=\"https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example.png\"><img loading=\"lazy\" width=\"1440\" height=\"384\" decoding=\"async\" src=\"https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example.png\" alt=\"an image showing human and AI generated text and how an AI rewrote each example\" class=\"wp-image-3139958\" srcset=\"https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example.png 1440w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-680x181.png 680w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-800x213.png 800w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-330x88.png 330w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-768x205.png 768w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-1030x275.png 1030w, https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example-1380x368.png 1380w\" sizes=\"(max-width: 1440px) 100vw, 1440px\"\/><\/a><figcaption class=\"wp-element-caption\"><span class=\"caption wp-caption-3139958\">A human and the AI model behind ChatGPT (GPT-3.5-Turbo) both wrote Yelp reviews, shown on the left. Researchers then fed those reviews back into GPT-3.5-Turbo. They told the AI model to rewrite them to be more concise. The revised reviews are shown on the right. Red text indicates what was deleted during revision. Yellow text marks what was added. The AI made a lot of edits to the Yelp review that the person wrote. But it changed almost nothing in the AI-generated review. <a href=\"https:\/\/www.snexplores.org\/wp-content\/uploads\/2024\/05\/1440_detecting_AI_content_example.png\">Click image to enlarge<\/a>.<\/span><span class=\"credit wp-credit-3139958\">Columbia University<\/span><\/figcaption><\/figure>\n<h4 class=\"wp-block-heading\">A red flag<\/h4>\n<p>Yang and Mao\u2019s team is working to make Raider an online tool that anyone can use. When it\u2019s done, people could send text through the tool and find out if it was likely AI-generated or not.<\/p>\n<p>Until then, the idea behind Raidar is easy for anyone to use, says Yang. \u201cYou don\u2019t have to be a computer scientist or data scientist.\u201d For example, a suspicious teacher could ask any chatbot to rewrite a student\u2019s work. If the bot makes very few edits, that could be a red flag that the student may have used AI.<\/p>\n<p>Bhattacharjee notes that teachers shouldn\u2019t take action based on Raidar\u2019s output alone. \u201cThe final judgment should not be based entirely on this tool,\u201d she says. That\u2019s because Raidar isn\u2019t always correct. Also, some students may have good reasons to use AI. For example, AI can help clean up grammar.<\/p>\n<p>Meanwhile, Yang is thinking about how something like Raidar might flag other types of AI-generated media. He\u2019s now studying what happens if you ask an AI model to revise an image, video or audio clip. If the model makes a lot of edits, that could indicate original human work.\u00a0<\/p>\n<div class=\"wp-block-group cheat-sheet-cta is-layout-flow\">\n<div class=\"wp-block-group__inner-container\">\n<h2 class=\"wp-block-heading has-text-align-center\">Do you have a science question? We can help!<\/h2>\n<p class=\"has-text-align-center\"><a href=\"https:\/\/forms.gle\/YbhPosFTMqjbSNnV7\" target=\"_blank\" rel=\"noreferrer noopener\">Submit your question here<\/a>, and we might answer it an upcoming issue of\u00a0<em>Science News Explores<\/em><\/p>\n<\/div>\n<\/div><\/div>\n<p><br \/>\n<br \/><a href=\"https:\/\/www.snexplores.org\/article\/a-new-test-could-help-weed-out-ai-generated-text\">Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Imagine that you\u2019re helping judge a writing contest at your school. You want to make sure everyone did their own work. If someone used an artificial intelligence, or AI, model such as ChatGPT to write an entry, that shouldn\u2019t count. But how can you tell whether something was written by AI? New research reveals a [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":61318,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"tdm_status":"","tdm_grid_status":"","fifu_image_url":"https:\/\/www.snexplores.org\/wp-content\/themes\/sciencenews-sns-child\/client\/src\/images\/cta-module-sm@2x.png","fifu_image_alt":"","footnotes":""},"categories":[606],"tags":[5533,848,6693,47203],"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/posts\/61317"}],"collection":[{"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/comments?post=61317"}],"version-history":[{"count":1,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/posts\/61317\/revisions"}],"predecessor-version":[{"id":61319,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/posts\/61317\/revisions\/61319"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/media\/61318"}],"wp:attachment":[{"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/media?parent=61317"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/categories?post=61317"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/news.talkwithrattan.com\/index.php\/wp-json\/wp\/v2\/tags?post=61317"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}