Using AI Responsibly in Teaching and Learning

This arti­cle presents the“Guide­lines for Uni­ver­sity Fac­ulty on the Use of Arti­fi­cial Intel­li­gence,” which pro­vide a con­cise and acces­si­ble sum­mary of the legal require­ments for using AI in every­day uni­ver­sity life.

Background

The rec­om­men­da­tions address ques­tions such as:

• Why is it impor­tant to address the topic of AI in teach­ing and exams?
• What skills do I need as an edu­ca­tor, and what skills should I teach my stu­dents?
• What are the legal lim­its on the use of AI, and who sets them?
• What should be con­sid­ered when using AI in exams, as well as in prepa­ra­tion for and fol­low-up to them?

The rec­om­men­da­tions were devel­oped as part of the project “AI in Stud­ies, Teach­ing, and Exams” by the Lower Sax­ony Dig­i­tal Learn­ing Hub (DLHN), which, as part of the Hochschule.digital Lower Sax­ony (HdN) was secured and funded by the Lower Sax­ony Min­istry of Sci­ence and Higher Edu­ca­tion (MWK) (future.lower-saxony) is funded. Fac­ulty mem­bers from var­i­ous uni­ver­si­ties and aca­d­e­mic dis­ci­plines in Lower Sax­ony con­tributed their exper­tise to dis­till com­plex issues sur­round­ing the use of gen­er­a­tive AI in teach­ing, learn­ing, and test­ing con­texts into prac­ti­cal, core mes­sages. The project’s goals include empow­er­ing fac­ulty and stu­dents to develop com­pe­tence in work­ing with AI, pro­mot­ing inno­v­a­tive teach­ing and assess­ment for­mats, fos­ter­ing sus­tain­able net­work­ing among uni­ver­si­ties in Lower Sax­ony, as well as Pro­vi­sion of OER for long-term sup­port in the form of work­ing mate­ri­als, posi­tion state­ments, etc. The rec­om­men­da­tions for action were also pub­lished under a CC0 license to make it as easy as pos­si­ble to reuse the con­tent for var­i­ous pur­poses.

Promoting AI Literacy

If stu­dents are to be allowed to use gen­er­a­tive AI in courses or exams, instruc­tors should first reflect on the extent of their own AI exper­tise. This includes, among other things, knowl­edge of legal reg­u­la­tions and inter­nal guide­lines of the uni­ver­sity, the depart­ment, the insti­tute, or the aca­d­e­mic dis­ci­pline. If nec­es­sary, new knowl­edge and skills may be required, which can be acquired through infor­ma­tion and pro­fes­sional devel­op­ment opportunities—such as those found via twillo’s “AI Get­ting Started” resource or the event cal­en­dar of the Net­work of State Insti­tu­tions for Dig­i­tal Higher Edu­ca­tion (NeL). In this way, instruc­tors are best able to ensure that stu­dents can develop the AI com­pe­ten­cies nec­es­sary for the tasks assigned to them.

In addi­tion to com­ply­ing with legal and insti­tu­tional require­ments, AI lit­er­acy can also help coun­ter­act the ero­sion of exist­ing skills—for exam­ple, through crit­i­cal reflec­tion on one’s own usage behav­ior and the con­se­quences of out­sourc­ing too many tasks to gen­er­a­tive AI sys­tems. The rec­om­men­da­tions for action pro­vide guid­ance on devel­op­ing AI lit­er­acy as well as ref­er­ences to infor­ma­tional mate­ri­als from the net­work of uni­ver­si­ties in Lower Sax­ony.

Eth­i­cal aspects of AI use are briefly addressed in the rec­om­men­da­tions; a more detailed dis­cus­sion can be found in the arti­cle “Ethics of AI Use in a Higher Edu­ca­tion Con­text.”

Building Trust

To alle­vi­ate stu­dents’ con­cerns about unin­ten­tion­ally using AI in an imper­mis­si­ble man­ner or vio­lat­ing their duty of care—but also to put instruc­tors in a bet­ter posi­tion should an attempt at cheat­ing occur—clear rules regard­ing the use of gen­er­a­tive AI should be estab­lished. Which AI-based tools are per­mit­ted, to what extent, and how should their use be doc­u­mented? As an instruc­tor, what do I expect in terms of reflec­tion, inde­pen­dent work, and acknowl­edg­ment of the use of these tools?

The rec­om­men­da­tions can serve as a guide for devel­op­ing usage poli­cies and guide­lines. They can also be used as a con­cise ref­er­ence work on legal prin­ci­ples to quickly gain an under­stand­ing of the use of gen­er­a­tive AI in courses and exams.

Require transparency

Sci­en­tific stan­dards include, among other things, the clear attri­bu­tion of oth­ers’ con­tri­bu­tions to one’s own work in order to high­light one’s own orig­i­nal con­tri­bu­tion in the analy­sis of sources and con­text. The EU’s AI Reg­u­la­tion also con­tains clear guide­lines for label­ing con­tent gen­er­ated by AI and con­tent cre­ated with the aid of AI. For exam­ple, image, video, and audio mate­r­ial must be labeled in an acces­si­ble and clearly rec­og­niz­able man­ner, unless it has already been labeled by the provider. Text must also be labeled in a clearly rec­og­niz­able man­ner, unless

• it has already been marked by the provider, or
• edi­to­r­ial respon­si­bil­ity is assumed, or
• it was final­ized in-house, or
• it is for inter­nal use only.

The arti­cle “What AI-Gen­er­ated Con­tent Do I Need to Label, and How?” also addresses dis­clo­sure require­ments regard­ing AI-gen­er­ated or AI-manip­u­lated con­tent.

It is not easy to ver­ify com­pli­ance with stan­dards for label­ing con­tent gen­er­ated by or cre­ated with the help of AI. How­ever, tech­ni­cal markings—such as dig­i­tal watermarks—or the use of AI detec­tors are not reli­able tools in cases where decep­tion is sus­pected. On the one hand, detec­tion can be cir­cum­vented rel­a­tively eas­ily; on the other hand, detec­tors can also pro­duce false-pos­i­tive results. A state­ment on the use of AI detec­tors to eval­u­ate exam per­for­mance, which was devel­oped as part of the DLHN ini­tia­tive, there­fore rec­om­mends instead the devel­op­ment of a new exam­i­na­tion cul­ture and cor­re­spond­ing exam­i­na­tion for­mats.

Strengthening Responsibility

By proac­tively address­ing ques­tions about the use of gen­er­a­tive AI in classes and exams, instruc­tors can help stu­dents under­stand the impor­tance of com­pe­tence, integrity, and trans­parency when work­ing with AI in a prac­ti­cal way and, ide­ally, pre­vent attempts at cheat­ing. Pro­mot­ing reflec­tive usage prac­tices, com­bined with clear guide­lines on per­mit­ted AI-based tools, the per­mit­ted scope of use, and the required attri­bu­tion and cita­tion can help stu­dents inter­nal­ize respon­si­ble aca­d­e­mic prac­tice within the frame­work of legal require­ments and treat exter­nal aids of any kind as such.

Bibliography

Miriam Bur­feind, Laura Fiegen­baum, Tom Hart­mann, Janine Horn, Julius Hoyer, Jas­min de Nys, and Maren Stephan (2026): Rec­om­men­da­tions for Col­lege Instruc­tors on the Use of Arti­fi­cial Intel­li­gence. DOI: https://doi.org/10.57961/rdsz-7320