In the next few blog posts, we will be look\u00ading specif\u00adi\u00adcally at the topic of AI in uni\u00adver\u00adsity teach\u00ading<\/strong>. The first ques\u00adtion will be to what extent tools for AI recog\u00adni\u00adtion<\/strong><\/em> should be used in exam\u00adi\u00adna\u00adtion papers. In the sec\u00adond arti\u00adcle, we will look at the legal and didac\u00adtic use of AI in the cre\u00adation of OER<\/a><\/em> . We then dis\u00adcuss prompt tips for high-qual\u00adity OER<\/em> and con\u00adclude the series with AI skills<\/em> that will soon be manda\u00adtory for uni\u00adver\u00adsity employ\u00adees. <\/p>\n\n\n\n The debate sur\u00adround\u00ading the use of arti\u00adfi\u00adcial intel\u00adli\u00adgence (AI) in aca\u00add\u00ade\u00admic exam\u00adi\u00adna\u00adtions is becom\u00ading increas\u00adingly impor\u00adtant. It is feared that unsu\u00adper\u00advised writ\u00adten work is more sus\u00adcep\u00adti\u00adble to cheat\u00ading and there\u00adfore can\u00adnot be retained as a form of exam\u00adi\u00adna\u00adtion. In par\u00adtic\u00adu\u00adlar, the ques\u00adtion of whether and how AI-sup\u00adported per\u00adfor\u00admance can be rec\u00adog\u00adnized or sanc\u00adtioned is of con\u00adcern to uni\u00adver\u00adsi\u00adties and courts alike. <\/p>\n\n\n\n The Admin\u00adis\u00adtra\u00adtive Court of Munich, for exam\u00adple, ruled on two appli\u00adca\u00adtions(appli\u00adca\u00adtion 1<\/a>, appli\u00adca\u00adtion 2<\/a>) from rejected appli\u00adcants to the Mas\u00adter\u2019s degree pro\u00adgram for pro\u00advi\u00adsional admis\u00adsion to the pro\u00adgram. They are said to have sub\u00admit\u00adted an essay for their appli\u00adca\u00adtion that was allegedly gen\u00ader\u00adated using AI. Both appli\u00adca\u00adtions were rejected by the court on the basis of prima facie evi\u00addence of decep\u00adtion. <\/p>\n\n\n\n Accord\u00ading to the prin\u00adci\u00adple of per\u00adfor\u00admance under higher edu\u00adca\u00adtion law, exam\u00adi\u00adna\u00adtions must be com\u00adpleted inde\u00adpen\u00addently and with\u00adout unau\u00adtho\u00adrized aids<\/em>. Can\u00addi\u00addates usu\u00adally sub\u00admit a cor\u00adre\u00adspond\u00ading dec\u00adla\u00adra\u00adtion of inde\u00adpen\u00addence. Accord\u00ading to the court, if this includes an explicit waiver of the use of AI, this con\u00adsti\u00adtutes a breach of reg\u00adu\u00adla\u00adtions if the work was cre\u00adated in whole or in part using AI. So-called AI detec\u00adtors can be used to deter\u00admine this. These are tools that ana\u00adlyze con\u00adtent to deter\u00admine whether it was gen\u00ader\u00adated by an AI. How\u00adever, the results of AI detec\u00adtors alone are not a deci\u00adsive indi\u00adca\u00adtion. An inde\u00adpen\u00addent opin\u00adion from human review\u00aders is also required, the court con\u00adtin\u00adued. <\/p>\n\n\n\n The court appar\u00adently assumed that the cur\u00adrently avail\u00adable detec\u00adtors are nei\u00adther accu\u00adrate nor reli\u00adable. In one study<\/a>, 14 detec\u00adtors were tested. It was found that the detec\u00adtors tended to clas\u00adsify the ana\u00adlyzed texts as writ\u00adten by humans rather than rec\u00adog\u00adniz\u00ading them as gen\u00ader\u00adated text. For texts that were cre\u00adated by an AI and then revised, the accu\u00adracy is said to be only 50 per\u00adcent. An AI-gen\u00ader\u00adated text that has been slightly<\/em> revised by an exam\u00adi\u00adnee may no longer be reli\u00adably rec\u00adog\u00adnized by the detec\u00adtors. <\/p>\n\n\n\n Due to these uncer\u00adtain\u00adties, proof of decep\u00adtion should be based on a human analy\u00adsis of typ\u00adi\u00adcal AI char\u00adac\u00adter\u00adis\u00adtics<\/em> in addi\u00adtion to the use of detec\u00adtors. These include texts with per\u00adfect form, with\u00adout spelling, punc\u00adtu\u00ada\u00adtion or gram\u00admat\u00adi\u00adcal errors, as well as typ\u00adi\u00adcal AI errors such as hal\u00adlu\u00adci\u00adna\u00adtions, exag\u00adger\u00ada\u00adtions or inac\u00adcu\u00adrate ref\u00ader\u00adences in quo\u00adta\u00adtions. Fur\u00adther infor\u00adma\u00adtion and tips on this can be found in the arti\u00adcle \u201cRec\u00adog\u00adniz\u00ading AI text in exam\u00adi\u00adna\u00adtion papers<\/a>\u201d by Matthis Kepser. <\/p>\n\n\n\n If you would like to learn more about the con\u00adnec\u00adtion between AI and OER, we would like to invite you to our next work\u00adshop<\/strong> \u201cAI and OER in use<\/a>\u201d. We will inform you about the legal aspects of cre\u00adat\u00ading open edu\u00adca\u00adtional mate\u00adri\u00adals with the help of AI<\/em>. We will also show you how AI can help to make OER more effec\u00adtive and reduce the work\u00adload. You are wel\u00adcome to reg\u00adis\u00adter at support.twillo@tib.eu<\/a><\/strong> to reg\u00adis\u00adter.<\/p>\n\n\n\n <\/p>\n\n\n\n Sam\u00adple tem\u00adplate as PDF (Adobe Reader)<\/strong><\/a>, DocX (Microsoft)<\/strong><\/a> or Open\u00adDoc\u00adu\u00adment (ODT)<\/strong><\/a> down\u00adload.<\/p>\n\n\n\n <\/p>\n","protected":false},"excerpt":{"rendered":" In the next few blog posts, we will be look\u00ading specif\u00adi\u00adcally at the topic of AI in uni\u00adver\u00adsity teach\u00ading. The first ques\u00adtion will be to what extent tools for AI recog\u00adni\u00adtion should be used in exam\u00adi\u00adna\u00adtion papers. In the sec\u00adond arti\u00adcle, we will look at the legal and didac\u00adtic use of AI in the cre\u00adation [\u2026]<\/p>\n","protected":false},"author":1,"featured_media":2986,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"wp_typography_post_enhancements_disabled":false,"footnotes":""},"categories":[145],"tags":[],"class_list":["post-2997","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-legal-issues"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/posts\/2997","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/comments?post=2997"}],"version-history":[{"count":0,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/posts\/2997\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/media\/2986"}],"wp:attachment":[{"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/media?parent=2997"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/categories?post=2997"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.twillo.de\/en\/wp-json\/wp\/v2\/tags?post=2997"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Examination papers with AI<\/h2>\n\n\n\n
The reliability of AI detectors<\/h2>\n\n\n\n
Download<\/h2>\n\n\n\n