Artificial intelligence in your studies and learning
ChatGPT, Google Bard, DeepL, and other applications of artificial intelligence (AI) available online are based on large language models. In recent years, language models have advanced to the point where they can produce human-like text and conversations. They can also improve and transform text at such a high level that it can be difficult to distinguish the final result from text generated by humans. In the future, more such models will undoubtedly emerge, and their capabilities will continue to evolve. It is therefore important that we consider their existence in studies, education, and research.
Since artificial intelligence provides new possibilities to produce text with unclear origins and reliability, it is important that we use them in a controlled manner. For example, teachers can restrict the use of AI-based tools in situations where their use would not enhance your learning. On this page, we provide concrete examples of how AI-based tools may and may not be used in studies. These examples are based on Hanken's guidelines for the use of artificial intelligence in teaching and learning. These guidelines are updated continuously as technology and AI-related legislation evolve. Therefore, we recommend visiting this page regularly to check for updates.
Hanken encourages teachers to use AI in their teaching where it is appropriate. This way, we can prepare you for a future society and work life where AI methods will be extensively used. If you are unsure whether you can use AI as support to complete a task, you can always ask the teacher before you begin working on your task.
By AI-based services, we refer to technological tools that can perform certain tasks equivalent to humans, such as identifying patterns in language, recognizing images, solving problems, and learning. This type of technology can, for example, assist you in:
- Rewriting something in a different way (paraphrasing).
- Improving grammar and spelling in your texts.
- Generating text and images based on guidelines you provide to the tool.
AI-based tools can serve as support and assist you in your learning, but they must be used correctly, and they do not replace your own competence!
In the workplace, AI-based tools offer solutions for tasks such as automating various routine tasks, generating data-driven insights in operations, diagnosing problems and inefficiencies, and forming initial ideas for new projects. It is highly likely that you will encounter and work with various AI-based tools during your career, so it is good to familiarize yourself with their use already during your studies.
AI-based tools can act as a tutor or coach, helping you build your own competence when studying independently. For example:
- You can use AI-based tools to create questions that help you check if you understand difficult concepts correctly. AI-based tools can also provide alternative explanations that may be easier to understand. However, remember to verify the accuracy of the information provided by AI!
- You can use AI-based tools to detect mistakes or language errors in your work or correct coding errors. By studying the errors flagged by AI, you can learn from them. You always have responsibility for the final text, so it is worth reviewing the changes suggested by AI before accepting the edited text.
- You can use AI-based tools to get ideas or a template for your own work. You can also receive feedback on your existing text and suggestions on how to improve the tone, structure, and argumentation of the text.
Hanken has specified cases where AI-based tools are not allowed. In addition to these cases, teachers may forbid the use of AI-based tools in a course when it prevents the completion of the course's competency goals.
If the use of AI-based tools is forbiden in a course, this should be stated in the detailed syllabus document on the course page in Moodle.
The situations and use cases that Hanken has forbidden include:
- All situations where the use of AI-based tools prevents completion of the course's competency goals, such as in courses where students are expected to create their own text and argumentation.
- Theses (applies to bachelor's and master's theses as a whole, and the main body of doctoral theses): It is forbidden to directly copy text created by AI-based tools or large language models and present it as your own. However, you are allowed to work with AI-based tools during the thesis process. Read more about AI and theses below.
- Maturity tests.
If the use of AI-generated text is discovered where it is not allowed, this constitutes cheating. Cheating is dealt with according to Hanken's action plan for Academic Dishonesty.
AI-based tools can be used during the thesis process as long as you follow the following two basic principles:
- You must always clearly explain how you have used AI-based tools in your work.
- You must never directly copy and claim as your own any text created by AI-based tools.
If you do not follow these two rules, your thesis may be failed. A failed thesis means that you must start over with a new topic. You cannot continue with the same research question.
How can AI-based tools be used during the thesis process?
AI-based tools can be useful in explaining the relationships between two different phenomena, a phenomenon and a theory, or different theoretical concepts. Especially when the keywords you want to learn about are very basic and stable, AI-based tools can generate clear explanations. You are allowed to use AI-based tools to make inquiries and to get an initial understanding of a new area or to test whether a real-world phenomenon appears relevant to a specific theoretical perspective (or vice versa). However, note that there is no guarantee that the answer generated by large language models is true, and that AI-based tools can produce fabricated references and "facts". You must verify the ideas suggested by AI-based tools and find sources that confirm them.
You cannot copy and include text answers generated by AI-based tools in your thesis*, nor can you reference such an answer as a source. AI-based tools do not understand the answers they have generated. These tools calculate the probability of specific words appearing next to each other based on massive data. This can lead to incorrect, misleading, or biased information. AI cannot flag these problems on its own; humans must review the text and find reliable references.
*The only exception is cases where you specifically need to show that AI has provided such an answer. In that case, you should put the text in quotation marks, add a reference to the tool you used, and explain exactly how you obtained that answer.
AI-based tools can assist you in processing your data and performing analysis. We update the information about different services as they become available. Just like with any tools you have used in data collection and analysis, you should describe how you have used AI-based tools in the methodology section of your thesis. Note that you are responsible for research ethics in your thesis process. This means that you must account for all the steps you have taken in data collection and processing. It also means that you are not allowed to input data into AI-based services that Hanken has not approved for this purpose.
Allowed ways of processing collected data (updated May 2023):
- Transcribing recorded interviews using the automatic captioning service in Panopto. (Click here for instructions)
Forbidden ways of processing collected data (updated May 2023):
- Uploading data to external services such as ChatGPT to have it identify patterns, recurring themes, or other information from the data. Large language models like ChatGPT store the information users upload to the service. This information is used to train the language model, and the model can use the uploaded data in responses to other users. In other words, information uploaded to publicly accessible large language models is not kept confidential. This is particularly important when dealing with sensitive data, but it is best to always follow this rule to avoid violating Hanken's data management process and Finnish data protection laws by mistake.
Formulating, editing, and structuring your text
AI-based tools can assist in formulating, editing, and structuring your text. However, note that the argumentation must be your own, and you can only use AI-based tools to improve your own text. Carefully review the proposed improvements offered by AI-based tools to identify possible errors and correct them. Also, ensure that the references in the text have not been changed and still match the original source.
If you use an AI-based service to rewrite a paragraph from a source in your own words (paraphrasing), you must be extra careful to ensure that the content is sufficiently different to avoid being mistaken for plagiarism. You cannot rely on suggestions provided by AI-based tools being of sufficient quality. You must carefully review the text so that you can stand behind the text you incorporate into your thesis and ensure that the citation to the original source is correct. You are always fully responsible for the final version you submit. Read more about how to reference the original source when referring to a longer passage from a source in Hanken's referencing guide.
Remember! When you have used large language models or other AI-based tools, you must describe this in detail in your work.
How to describe the use of AI-based tools is explained in the next section.
The basic rules for the use of AI-based tools at Hanken are that the writer should always describe in detail how the tool has been used in the work. This applies to both smaller coursework and dissertations. Simply referencing ChatGPT is not sufficient; the description should also include information on exactly how the AI-based tool has been utilized. The teacher can provide specific instructions regarding this.
The description should be placed where it is appropriate to discuss how you have worked during the writing process, such as in the context of describing the structure of the work or in the methodology chapter. The specific location depends on the type of work. If you are unsure, it is advisable to consult the course instructor or your supervisor.
|Use case||Description in writing||Citation in the list of references|
ChatGPT has been used to generate ideas and a template for course work.
"I have used ChatGPT to generate ideas for important factors to consider in a PESTEL analysis with the question "conduct a PESTEL analysis for the logistics industry," to which ChatGPT generated the response "PESTEL is an environmental analysis that considers six factors: political, economical, social, technological, ecological, and legal factors. In the logistics industry..." (OpenAI, 2023).
If the response is relatively short, you can paste the entire answer. However, for longer responses, it is better to include them in an appendix to your work. The description should be followed by a reference to the creator of the AI-based tool; for ChatGPT, it is the company OpenAI. In the reference list, each prompt or question posed to AI-based tools should be listed as a separate reference.
OpenAI. (2023). ChatGPT (version May 3) [Large language model]. https://chat.openai.com/chat
Please make a note of which version of ChatGPT has been used. OpenAI's ChatGPT uses dates in their versions but other tools might use other ways, like numbers, to express the version, e.g. version 3.6.9.
Automatic subtitle service in Panopto has been used to transcribe interviews.
|"The interviews were transcribed using an AI-based service in the Panopto program. The program created an initial version of the transcript that I then reviewed, correcting misinterpreted words and adding information about the speaker."||No reference needed|
|ChatGPT has been used for improving spelling and language errors, wordings and structuring of the work.||
"To improve the language and structure in my work, I have used an artificial intelligence language model called ChatGPT (OpenAI, 2023).
The work underwent a two-step improvement process using ChatGPT. First, I inputted my original text into the model with the purpose of reviewing grammar, vocabulary, and coherence to enhance the overall quality. Suggestions from ChatGPT were carefully reviewed and used where appropriate. Next, the structure of the work was improved through iterative interactions with ChatGPT. I provided guidelines and specific prompts related to the organization and flow of ideas in my original text, based on which ChatGPT generated suggestions and alternative phrasings to clarify the logical coherence and argumentative structure of my work. I went through all the suggestions and only incorporated those into the final version that maintained the original intention and voice of my work.
It is important to note that the role of ChatGPT was strictly limited to providing suggestions and assisting in the revision process. I made the final decisions regarding the use of suggestions and revisions to ensure that the work reflects my perspective and understanding of the subject."
Please note that this is an example description. Your description should accurately reflect what you have done.
|OpenAI. (2023). ChatGPT (version May 3) [Large language model]. https://chat.openai.com/chat|
Instructions for the citation of AI-based tools are updated when necessary. Read more on APA's webpage.
Teachers are allowed to use detection programs to identify AI-generated text. If you have not clearly expressed how you have utilized AI in your work, you run the risk of being accused of cheating if AI-generated text is detected in your work. False positives are also possible. To safeguard yourself against accusations of cheating, it is important to be able to provide evidence of the steps you have taken in the writing process. Being able to openly account for the steps taken in the process is a central aspect of academic integrity and reliability. So, this is not something new, but due to the possibility of false positives from detection programs, more students may be required to prove the steps they have taken in the writing process.
- Save all draft versions of your work until your work is approved and assessed. For example, everytime you start writing, start by saving your work with the current date in the filename. This way, you can demonstrate the progress of the writing process retrospectively.
- Keep raw data until your work is approved and assessed. Remember to follow Hanken's data management process!
- Always describe openly, clearly, and in detail how you have used AI-based tools.
Use common sense! Consider first whether using AI-based tools is ethically right, responsible, and beneficial for your learning in the task. If you want to use AI-based tools to avoid doing something that is a central part of the task, it constitutes cheating.
All cases of cheating are handled according to Hanken's action plan for Academic Dishonesty.
Fabrication of data, references, and facts always constitutes cheating, even if you are not aware that the answer provided by the AI-based tool is fabricated. You are responsible for the work you submit. You must be aware of all references and verify their reliability.