TLU Recommendations for the Use of Artificial Intelligence, as of 08.05.2023

1.1. Create academic, literary and legal texts, as well as non-literary texts for daily use (ChatGPT, Jasper, WriteSonic, Notion AI, Wordtune,
Scholarchy, Adept and Summate.it);
1.2. Create visual material, including films, videos etc. (Dall-E, StableDiffusion, Midjourney, etc.)
1.3. Generate audio files, including music (MuseNet, AmperMusic, etc.);
1.4. Write code / programme (CoPilot, ChatGPT etc.);
1.5. Learn a new language (Duolingo, El Lingo etc.)

2.1. It will always require human input (i.e. there has to be a human who gives a specific command to do something in the required manner);
2.2. The sources of texts created with the help of Chat GPT and other text generators cannot be checked as the creation of texts is based on the statistical probability of the sequence of words;
2.3. References used in the texts created with the help of AI can be fictitious /wrong;
2.4. There are paid, free and combined services on the market. Paid versions of AI are often faster and provide better quality as compared to free services. Access to free versions might be more complicated, for example, during the examination period, at the time when there are more users, when a new/improved versions has been launched to the market, etc.;
2.5. Texts and visual materials produced with the help of AI can have issues with bias related to gender, race, language etc. (due to materials found on internet on the basis of which the models have been trained):
2.6. There could be issues with the user privacy and information security (i.e. it is not known which user data is collected, who processes the data and how).

3.1. Within the context of widespread use of AI, it is of crucial importance to discuss the topics of reliability of science with the students, including the importance of evidence-based methods of creating knowledge.
3.2. For the purpose of developing students’ digital and research- related competences, it would be advisable to test during studies the contexts in which the AI software appropriate for a specific field of studies works and the results it provides (for example, in the phase of generating ideas, text processing phase, replacing routine tasks etc.). The risks referred to in 2.5. and 2.6. should be considered.
3.3. When choosing AI software, it is important to consider that all students should have equal access to it (see 2.4.).
3.4. When integrating topics/activities related to AI into studies, good academic practices should be taken as the basis (including TLU Good Practice of Teaching and Supervising, from the viewpoint of students, the Good Practice of Learning is also important).
3.5. A member of teaching staff may allow or restrict the use of AI in his/her subject if this may impact the academic side and the achievement of learning outcomes.
3.6. At the beginning of the course, members of teaching staff should inform the students of the principles applied in their subject. Information about the impact of the use of AI on grading should be provided in the course description and should be made available to students in the Study Information System at the beginning of the course. The use of AI is considered as permitted unless it has been clearly banned by the member of teaching staff.
3.7. Students are obliged to follow the principles provided by the member of teaching staff. If the student has ignored the principles for creating work which is submitted for grading, it is considered academic fraud.
3.8. Members of academic staff may use AI when making preparations for teaching (e.g. adjusting study materials for students with special educational needs etc.) or when conducting studies (written feedback provided to students, etc.), but only after having checked and being convinced that the material created by AI is correct and informing students that AI has been used.

4.1. The author of the work is fully responsible for the content and form of the presented work (i.e. correctness of the information, citing etc.) even if the work or parts thereof have been created with the help of AI.
4.2. The author of the work is responsible for making sure that the work created with the help of AI does not violate the rights of other authors (including texts, photos, works of art, etc. available on the internet) or the principles of personal data protection (e.g. if necessary, permission to use and publish material depicting or related to a concrete person must be sought).
4.3. AI must not be used to fabricate data/information or to manipulate data/falsify information.
4.4. When using AI, a reference should be made to which program was used, how and in which part of the work.

APA citing guidelines are available at: https://apastyle.apa.org/blog/how-to-cite-chatgpt

MLA handbook: https://style.mla.org/citing-generative-ai/.

Chicago manual:https://www.chicagomanualofstyle.org/qanda/data/faq/topics/Documentation/faq0422.html.

1. Sources of inspiration of how other universities have regulated the use of AI:
   1.1. The University of Helsinki instructions to academic staff and students. 06.03.2023. https://studies.helsinki.fi/instructions/article/using-ai-support- learning. Seen 03.05.2023.
   1.2. The University of Konstanz recommendations to academic staff for spring semester 2023. https://www.uni-konstanz.de/en/lehren/weiterentwicklung-der-lehre/ki-in-der-lehre/. Seen 03.05.2023.
   1.3. The University of Copenhagen newsletter on the university website. Camilla Skovgaard Thomsen, The future is now: UCPH softens up on AI rules. 23.04.2023. https://uniavisen.dk/en/the-future-is-now-ucph-softens-up-on-ai-rules/. Seen 03.05.2023.
   1.4. Guidelines developed by the University of Tartu for using a text robot in studies, on the university website (in Estonian): https://ut.ee/et/sisu/tartu-ulikooli-suunis-tekstiroboti-kasutamiseks-oppetoos. Seen 03.05.2023.

2. Official guidelines:

2.1. European Commission, Directorate-General for Education, Youth, Sport and Culture, Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for Educators, Publication Office of the European Union, 2022, https://education.ec.europa.eu/news/ethical-guidelines-on-the-use-of-artificial-intelligence-and-data-in-teaching-and-learning-for-educators.

2.2. Ministry of Education and Research, ChatGPT ja kool: suuniseid tekstirobotite kasutamiseks hariduses (in Estonian).
https://www.hm.ee/sites/default/files/documents/2023-03/ChatGPT%20ja%20kool%20UUS_1.pdf.

3. Bias in AI:

Schwartz, Reva; Vassilev, Apostol; Greene, Kristen; Perine, Lori; Burt, Andrew; Hall, Patrick. Towards a Standard for Identifying and Managing Bias in Artificial Intelligence. National Institute of Standards and Technology Special Publication 1270 (2022). https://doi.org/10.6028/NIST.SP.1270. Seen 03.05.2023.

4. How does AI change learning, teaching, creation:

4.1. Hint, Helen; Lemendik, Helena; Leijen, Djuddah. ChatGPT kõrghariduse voodi all - kas päris koll?(in Estonian). Online portal of Estonian Public Broadcasting (ERR) Novaator, 03.05.2023. https://novaator.err.ee/1608966095/chatgpt-korghariduse-voodi-all-kas-paris-koll. Seen 03.05.2023.

4.2. Karjus, Andres. ChatGPT and the new text generators. Overview of the workshops that took place at TLU. https://datafigure.eu/reports/chatgpt_eng.html. Seen 03.05.2023.
4.3. Laas, Oliver. ChatGPT ja tehisintellekti plagiaat (in Estonian). Online portal of Estonian Public Broadcasting (ERR), 11.01.2023.
https://www.err.ee/1608846064/oliver-laas-chatgpt-ja-tehisintellekti-plagiaat. Seen 03.05.2023.
4.4. Laas, Oliver. ChatGPT eetilisest kasutamisest koolides (in Estonian). Online portal of Estonian Public Broadcasting (ERR).
https://www.err.ee/1608922886/oliver-laas-chatgpt-eetilisest-kasutamisest-koolides. Seen 03.05.2023.
4.5. Laas, Oliver. Moraliseerivad tööriistad (in Estonian). Online portal of Estonian Public Broadcasting (ERR). 08.05.2023.
https://www.err.ee/1608970966/oliver-laas-moraliseerivad-tooriistad. Seen 08.05.2023.
4.6. Trasberg, Henrik. Tehisintellekt toob kasu, aga ka riskidega tuleb tegeleda (in Estonian). Online portal of Estonian Public Broadcasting (ERR). 06.04.2023. https://www.err.ee/1608939326/henrik-trasberg-tehisintellekt-toob-kasu-aga-ka-riskidega-tuleb-tegeleda. Seen 03.05.2023.

5. Is it possible to detect the use of AI in student work:

S. Eaton. The use of AI-detection tools in the Assessment of Student Work, 06.05.2023. https://drsaraheaton.wordpress.com/2023/05/06/the-use-of-ai-detection-tools-in-the-assessment-of-student-work/. Seen 08.05.2023.

Andres Karjus- TLU School of Humanities

Oliver Laas- TLU School of Humanities

Kersti Papson- academic affairs ofice