THE USE OF AI IN TEACHING AND RESEARCH

Artificial intelligence (AI) refers, in particular, to generative artificial intelligence that is based on a Large Language Model (i.e., a self-regulated neural network based on vast amounts of textual teaching material). Applications that use generative AI (e.g., ChatGPT, GPT-4, Midjourney, Dall-E, Runway AI, Microsoft Copilot) produce text, video, audio, code, or images through prompts.

Therefore, the term artificial intelligence application specifically refers to the aforementioned types of AI-based services, not to the use of AI features integrated into other software.

These guidelines do not apply to research on or the development of AI, but to the general use of these services in research and teaching/studies.

These guidelines will be applied as a general guideline at the University of Eastern Finland. Specifying instructions may be provided separately by faculties within the limits of these general guidelines and legislation.

The University of Eastern Finland encourages the use of AI in teaching and research, taking into account, however, the restrictions set out in these guidelines and due to legislation and various contractual obligations.

1) As a rule, AI applications can be used in teaching and to support writing. The use of AI applications in studies (e.g., exams, assignments, essays, seminar assignments) may be prohibited in the course description or by the teacher in charge at the beginning of the course in writing (e.g., in the learning environment) if there is a risk that the use of models might have a negative impact on achieving the set learning outcomes.

2) If an AI application is used in the production of a work that is submitted during a course, the application (e.g., ChatGPT) used and the manner in which it was used must be specified in writing. This also applies to academic theses. The research-related guidelines for the use of AI will be applied to theses. Theses must not be completed entirely by using an AI application. Each thesis must include a sufficient amount of independent work to ensure that the learning outcomes set for the thesis are achieved. AI must not be listed as the author of texts or other written outputs. AI cannot be responsible for the content of the text – the person who wrote it will be responsible for the accuracy of all material.

Clarification: The teacher will provide more information on how the use of an AI application should be reported. Typically, this is done by specifying, in connection with the study attainment (e.g., an assignment), which AI application was used and in which manner.

3) Students may only be required to use AI applications that are free of charge to them. AI applications may require the creation of a user ID. Such AI applications collect personal data.

Clarification: Situations where students are required to use an AI application may be related to, e.g., studying the use of AI.

Clarification: The use of some AI applications requires the creation of a user ID, for which reason it may be necessary to provide the application with personal data. This is problematic if the AI application collects personal data (a username that can be associated with a person, such as a name-specific e-mail address) and its servers are located outside the EU or the EEA. Such AI applications include ChatGPT. Remember to verify the type of the AI application.

4) No confidential information, such as study attainments (e.g., learning assignments, essays, exam answers and learning diaries) or research material (e.g., interview transcripts) must be entered into external AI applications (e.g., ChatGPT). If personal data is processed in an AI application, the requirements laid down in the EU Data Protection Regulation and the Data Protection Act must be complied with.

Clarification: No personal data of the students (direct or indirect identifiers or other identifiers to be interpreted as being personal data) must be exported to AI applications, except when the university has concluded an agreement with the provider of the said AI application that contains conditions related to the processing of personal data.

5) If a student uses an AI application during a course or its part or in an exam situation where the use of AI has been forbidden in advance, the student’s activities are deemed to be fraudulent and will be treated in the same way as other cases of fraudulent conduct. The same applies to the above situations, even if the student reports that they have used artificial intelligence. The use of AI applications can be investigated, e.g., by means of an electronic plagiarism identification system.

6) Where required, the faculty, unit or independent institute (e.g., the Language Centre) may issue supplementary guidelines on the use of AI in their own teaching.

1) Copyright and ownership of background material and results

AI applications may have different terms of use as regards copyrights. The application’s copyright terms and conditions must be reviewed before the material produced by AI is used, e.g., publicly or with partners. It must also be ensured that the conditions of the AI application allow for the planned use of the material.

The ownership of the background material and research proposals will always remain with the original owner / as specified in the terms and conditions of the research contract, regardless of the use of an AI application. However, instructions on secrecy must always be taken into account when using AI (background material of research as well as research proposals are confidential).

Clarification: The background material of research (e.g., interview transcriptions and questionnaire responses) may be owned either by the university or the researcher. If the material is owned by the university and is in the university’s possession, the Act on the Openness of Government Activities (621/1999), the Archives Act (831/1994) and the Information Management Act (Act on Information Management in Public Administration 906/2019) apply to the material. This means, e.g., that confidential material that is in the university’s possession cannot be processed in an AI application.

2) User rights

AI applications may require the application owner to be granted, e.g., a parallel right of access to results generated by AI. This and any conflicting conditions of the research contract or the funder must be taken into account when using AI and when material generated by AI is used in research.

3) Artificial intelligence in externally funded projects and the funder’s terms

If the copyright terms and conditions of the AI application are in conflict with the funder’s funding terms, this will prevent the use of the application in the project, unless the funder separately gives a prior written consent to the use of the AI application. For example, commissioned research requires that the university transfers all copyrights to the client. In this case, no copyrights must remain, e.g., with the AI application.

4) Processing of confidential information

No confidential information must be processed in an AI application. Confidential documents include, e.g., research proposals and article manuscripts (Act on the Openness of Government Activities 621/1999, section 24), background material of research and other material based on the university’s contractual obligations (commissioned research).

No information contained in contracts or information that must be kept secret based on contract terms must be exported to AI applications. When the university commits to agreements other than those based on the university’s own model agreement templates and general terms and conditions, the agreements must be submitted to the university’s legal services for comments.

5) Personal data

Research material that contains personal data (direct or indirect identifiers or other identifiers to be interpreted as being personal data) must not be exported to AI applications, except when the university has concluded an agreement with the provider of the said AI application that contains conditions related to the processing of personal data. In addition, research subjects must always be informed of the processing of their personal data by means of AI. If the material contains personal data, the analysis of the research data cannot be completed by using an AI application without conducting a data protection impact assessment (PIA/DPIA).

The use of some AI applications requires the creation of a user ID, for which reason it may be necessary to provide the application with personal data. Universities cannot require their researchers to create user IDs for AI applications that require the disclosure of personal data (e.g., an email address and contact information that can be linked to a person).

In unclear situations, the university’s Data Protection Officer must always be contacted.

6) Research ethics

The University of Eastern Finland is committed to following the Finnish Code of Conduct for Research Integrity and Procedures for Handling Alleged Violations of Research Integrity in Finland issued by the Finnish National Board on Research Integrity, TENK. The guidelines specify what responsible conduct of research is.

The use of AI in research must be mentioned when the results of the research are published. The use of AI in research must be mentioned in the documents produced during the life cycle of the research, such as the research proposal, press releases and research permit applications, as well as when the research results are published. In such a case, the application that was used in the research and the purpose for which it was used must be specified.

7) Publications and the instructions of the publisher
Different publications, publication platforms and publishers have varying guidelines and rules on the use of AI applications. These instructions must always be taken into account in the case of materials produced with the help of AI: Some publishers have, in their publication terms, restricted the option to use AI applications or, at the minimum, require unambiguous mention of the use of AI in the production of the material.