Hamburg University of Technology AI Policy

Students use AI tools

critically and construc­tively to handle tasks –

both individually and in

groups – and reflect on

the results generated.

Use of AI as inspiration

for structuring written

essays. Critical reflec­tion on an AI-gener­ated text based on

one’s own professional

expertise.

The examiner ultimately determines

the guidelines for the specific examination in ac­cordance with the ASPO and FSPO (cf. Spannagel,

2023).

In terms of examination procedures, AI tools do not di­rectly trigger the need for changes to examination reg­ulations. Rather, it is a pedagogic decision or rather a

decision to be made by lecturers and examiners as to

whether AI tools may be used when completing exami­nation achievements, and in what way, or whether their

use is excluded.

An achievement shall be

deemed to be that of an individual if it has been ren­dered without the use of unauthorised aids and without

unauthorized assistance from third parties. This means

that the candidate is obliged both to ensure that the

achievement in question is their own work and not to be

misleading about this.

Supervised examinations or those which require di­rect interaction with examiners – such as in-person

written examinations, oral or practical examinations –

are initially relatively unaffected by the new possibili­ties, as long as AI tools are not explicitly designated as

permitted aids for voluntary use.

A fundamentally open and responsible ap­proach to AI is therefore considered constructive.

Students work on a topic as part of a group and present it during the seminar. The stu­dents are tasked with:

a. Gaining an overview of the topic with support from AI tools;

Students ask subject­related, methodological

and AI-tool-related ques­tions in advance.

AI tools support the provi­sion and preparation of

materials (scripts, articles,

textbooks, explanatory

videos) + task creation

(sample code).

Students

… read code and deter­mine the output.

… develop a code and

discuss the result.

… discuss the voluntary

use of AI tools.

AI tools can suggest

exam questions, ru­brics and reflective

questions on AI usage

or NON-usage.

The lecturer provides students with the material and an assignment.

The students are tasked with:

a. Reading a code and determining the output;

b. Developing the code for a given problem and discussing the solution;

5. eine eidesstattliche Erklärung auf einem vom Promotionsausschuss

vorgeschriebenen Formblatt „Eidesstattliche Erklärung (PDF)” darüber, dass

a) die Arbeit selbständig angefertigt worden ist,

b) die örtlich oder inhaltlich aus anderen Quellen entnommenen Stellen

als solche kenntlich gemacht sind,

This use of

AI tools is considered part of everyday work and is therefore

not subject to acknowledgement requirements, unless ex­aminers explicitly request acknowledgement in line with

professional practice. It can then be noted in the Declara­tion of Originality (cf. Buck and Limburg, 2023) or in the list

of tools used (cf. Spannagel, 2023) that AI tools were used

or more specifically that the text in chapter XY or section

XY or line XY was rephrased or translated using AI tool Z

There is room for

interpretation in each individual case: Are AI-sup­ported tools part of everyday work (cf. Eaton, 2023)

and therefore do not have to be specified, or do they

go beyond this and concern more significantly the

key scientific competence to be tested? Accordingly,

“subject-specific examination regulations can de­fine to what extent and under what conditions the

use of AI-supported tools is permitted” (Hoeren,

2023, p. 23). The examiner ultimately determines

the guidelines for the specific examination in ac­cordance with the ASPO and FSPO (cf. Spannagel,

2023).

A requirement to identify any text output from an AI

writing tool can also result from the terms of use of the

respective software (cf. Hoeren, 2023, p. 23).

This use of

AI tools is considered part of everyday work and is therefore

not subject to acknowledgement requirements, unless ex­aminers explicitly request acknowledgement in line with

professional practice. It can then be noted in the Declara­tion of Originality (cf. Buck and Limburg, 2023) or in the list

of tools used (cf. Spannagel, 2023) that AI tools were used

or more specifically that the text in chapter XY or section

XY or line XY was rephrased or translated using AI tool Z

The methods examiners use to scrutinise students’ work

are of crucial relevance. With regard to assessing exam­ination achievements, it is also important to treat all

candidates equally and to apply a standardised exami­nation and assessment standard.

This means

that the candidate is obliged both to ensure that the

achievement in question is their own work and not to be

misleading about this.

Stellt sich nach Abschluss des Promotionsverfahrens heraus, dass der

Doktorgrad durch Täuschung oder auf sonstige unrechtmäßige Art und Weise

erworben worden ist, so spricht der Promotionsausschuss die Unwürdigkeit der

Promovierten oder des Promovierten aus. Der Doktorgrad ist dann zu entziehen, so

dass der Grad oder Titel nicht mehr geführt werden darf. Unter den Tatbestand der

Täuschung oder des unrechtmäßigen Erwerbs sind auch die Fälle zu subsumieren,

in denen jemand den Doktorgrad über Dritte erworben, Plagiate verwendet,

wissenschaftliche Ergebnisse gefälscht oder in anderer Weise im Zusammenhang

This guidance is primarily aimed at lecturers – but also

serves as a guide for students at TU Hamburg – to inte­grate new AI developments into teaching and learning

processes.

The aim is not to replace lecturers

with AI tools, but rather to support them in their

various roles by using AI tools.

In this respect, we consider the integration of AI tools

into teaching at TU Hamburg as a step-by-step process

that requires careful planning, design, implementation,

reflection and evaluation, taking into account pedagog­ical principles, university-specific framework conditions

and the expertise and responsibility of teaching staff.

AI tools support the provi­sion and preparation of

materials (scripts, articles,

textbooks, explanatory

videos) + task creation

(sample code).

AI tools can suggest

exam questions, ru­brics and reflective

questions on AI usage

or NON-usage.

As soon as personal data is processed using AI tools, the

basic principles of data protection law from Art. 5 of

the General Data Protection Regulation (GDPR) must be

taken into account. Furthermore, there must be a legal

basis for processing this personal data in each individ­ual case.

The use of AI soft­ware from an American company routinely fails to com­ply with European data protection requirements.

For this reason,

too, it is essential to ensure that data security re­quirements are followed when using AI tools and that

access by AI tools to sensitive, internal or confiden­tial information provided specifically for this pur­pose is strictly avoided.

This

function can be switched off on some tools. If the AI

tool used features the relevant settings option and

personal data is to be processed, deactivating the

training mode is recommended (privacy by default).

DSGVO-konform KI-Tool

Die selbst gehosteten Open-Source-Modelle laufen vollständig auf der HPC-Infrastruktur der GWDG in Göttingen. Nutzereingaben und Antworten werden werden dabei nicht gespeichert und nach der Verarbeitung sofort verworfen. Chat-Verläufe existieren nur lokal im Browser und werden beim Schließen der Sitzung gelöscht.

Bei Nutzung der externen OpenAI-Modelle werden die Anfragen an Microsoft-Server in europäischen Rechenzentren weitergeleitet – persönliche Informationen in den Anfragen werden dabei an den externen Dienstleister übermittelt.

In order to write this guidance, the Committee for

Strategy Development in Teaching and Learning set

up a working group incorporating all status groups

(professors, academic staff, technical and adminis­trative staff, and the student body), led by Prof. Ma­ren Baumhauer. This guidance was reviewed from a

legal perspective and discussed by the Executive

Board.

We therefore plan to treat the

guidance as a “living document” and regularly adapt

it to reflect our current level of understanding and

any new developments.

In line with our

guiding principle, we want to promote the development

and testing of AI tools in teaching at TU Hamburg in a

critical, constructive and technologically reflective

manner, thus pro-actively tackling associated teaching

tasks.

Eine zeitgemäße Hochschullehre berücksichtigt das Digitale nicht nur auf Ebene von

Lehr- und Lernformaten, in denen digitale Medien als Mittel zur Wissensvermittlung, zur

Kompetenzförderung oder zur Unterstützung von Kollaboration und Kommunikation

eingesetzt werden, sondern integriert Digitalität als umfassenden gesellschaftlichen

Veränderungsprozess auch in ihre Studieninhalte.