AI for Teaching: Legal Recommendations

Legal Recommendations

The general parts of bachelor's- and master's examination regulations,  in particular §8 and §9 AT, contain regulations which require a declaration of originality to be submitted for written work. This regulates the use of aids in preparing written work, which is also applicable to the use of AI-based tools. In addition, the regulatoins listed in § 7 DigiPrüfOapply for digital examinations.

AI-based systems and other tools may be permitted as aids, but there cannot be an obligation to use these.

However, rules must be defined by teaching staff and examiners and communicated accordingly. This applies in particular because students are always responsible for the results of their work and, in the case of the use of AI-based systems, it must be made clear what exactly is the work of the students and what was the result of the AI-based system's support. The task of explaining the scope and nature of the use of AI-based systems and the assumption of responsibility for the results lies with the creators.

Proof of use and documentation, including a list of "works cited", must be specified. Legitimate assessment requires that lecturers know what the person submitting work has achieved.

In order for the student's own work to be assessed, it must be clear in the source reference what is the student's own contribution was as well as what was the work of the AI-based systems. Teachers must communicate exactly what evidence is required.  For example, lecturers can request that the prompts and/or prompt sequences used are submitted or that the use of AI-based systems is documented transparently.

It is recommended that teaching staff members, degree programs, and faculties agree on a basic procedure for their degree programs. Particular attention should be paid to this:

• the quality of coursework and examinations in which the results of AI-based instruments are used is the responsibility of the students.
• the principles of good scientific practice always apply, and require students to be particularly transparent about AI use.
• copyright issues have not been conclusively clarified in legal terms; the current regulations must always be observed. If the data on which AI-based systems are based contains work that has been plagiarized, the answers generated by the tool could also be considered plagiarized. Anyone who reproduces or publishes such a text in accordance with Sections 16 and 19 of the German Copyright Act (UrhG) is committing a copyright infringement. In the examination performance, the lack of indication of the source would then be considered plagiarism and therefore treated in accordance with Section 18 AT.

When dealing with personal data of third parties in connection with AI-based systems, compliance with Data Protection Law, in particular the GSVO and the Bremen Implementation Act to the EU-General Data Peotection Regulation (BremDSGVOAG), must be ensured. This applies both to the upload of information to AI-based tools and to the content generated by them.

In suspected cases of attempted cheating or violations of good scientific practice, the usual examination procedure according to § 18 AT (attempted cheating) is to be implemented. This lists multiple types of unauthorized aids, which would also include AI.