Generative Artificial Intelligence and University Study

Given all the limitations of generative AI tools (in particular when it comes to accuracy, reliability, attribution, and accountability), it becomes important to approach them with a critical mindset. So, what role could generative AI play within the context of your University study? 

For non-assessed tasks and enhancing your learning, generative AI tools may offer some benefits. For example, an initial place to get basic explanations of concepts that can springboard deeper research; or a soundboard for generating ideas and potential options for projects; or an additional check for your spelling and grammar. 

If unsure whether to use Generative AI, think of the following questions:  

• What is my learning objective and could an AI tool help me develop as a learner? 

• Do I have sufficient knowledge and/or evidence to verify information and factcheck for inaccuracies? 

• Do I have clear reasons for the choices I make (e.g. the prompts I use)? 

• Is the AI helping with my understanding of concepts and ability to ask further questions? 

• Do I have sufficient agency and control over my interaction with the AI tool? 

When is it safe to use ChatGPT? (Source UNESCO 2023, p.6).

Remember:  

For assessed tasks, you should be given guidance in your assessment brief on whether use of GenAI is permitted. Assessments are classified as follows:

• Category 1= GenAI may not be used;
• Category 2 = GenAI may be used to support learning; and
• Category 3 = GenAI use is encouraged and incorporated in the assessment task.

Most 'traditional' coursework will fall under Category 2. If uncertain, do ask your module convenor for clarification.

When you are preparing your assignment, it is important that you maintain agency and control over the work you create. In the same way you would not be asking a friend to write your essay, you should not outsource your work to an AI chatbot! It may be okay to chat with an AI bot (like you would do with one of your colleagues) about potential topics to look into, about the key concepts of relevance, or about possible ways to structure your assignment. However, the decision making and responsibility lie with you. An AI chatbot cannot be trusted to provide accurate, thoroughly researched, properly credited, and thoughtful analysis – this is your job, and it is what will help you learn in the process.

Remember the reason why you've come to university and the value of your degree: you're here to develop your own knowledge and skills. 

To summarise:  

FAQs: 

• What if using Generative AI is part of the assignment instructions?  

• In some cases, it may be part of your assessed task to generate AI outputs and use them in various ways (these assessments will fall under Category 3); for example, you may be asked to critique them, to evaluate and select among different options, to assess biases that may be embedded in these, or to compare them with your own work. In any such case, you must follow the instructions given by your module convenor and use the AI tool at the appropriate stage of the task and for the specified purpose. You must also acknowledge and describe how the AI outputs were generated and used in your work. If, after checking the instructions, you have any questions about using AI as part of your assignment, check with your module convenor for clarification.  

• What if a software product I have been using has automatically added new AI features? Is it still okay to use this software? 

• AI-based features included in commonly used word-processing software may become more and more the norm. Such features may look like advanced autocomplete options, or may make personalised suggestions that you can adopt or reject. Normally, you do not have to acknowledge use of these features. Remember, however, to use their suggestions according to your own judgement. 

• It all sounds confusing! Why should I bother engaging with AI at all? 

• It is important to experiment with and understand these tools, as they are becoming part of our lives. As a student at university, you would be expected to decide when and how to use these tools, with criticality and accountability. In addition, making informed decisions about deployment, management, and use of AI tools in the workplace may become a key employability skill.  So, the answer is not to shy away from this technology, but to test it, in order to understand its limitations and come up with beneficial applications. 

Some examples of developmental use of Generative AI include:  

1. Asking AI to generate revision questions on a topic, which you can then use for practising in exam conditions. 

2. Brainstorming titles for the article you plan to publish based on your dissertation or thesis. 

3. Exploring alternative ways to structure your essay; you can then compare all the options, and decide which one you prefer (or come up with a new one!) 

4. Factchecking and testing its abilities and limits through experimentation. Discussing your findings with your peers and/or your tutors. 

Can you think of any other applications? 

Some examples of unsafe and/or unethical use of generative AI include: 

1. Asking the tool to produce answers, designs, or code that you then include in your assignment submission (regardless of whether use of AI is acknowledged), unless explicitly asked to do so. 

2. Asking for information and trusting the answer without independent verification and fact checking. 

3. Submitting something - e.g. a blog post, an article, an essay, an image - that was produced using AI, without acknowledgment of how it was generated. 

4. Entering in a Generative AI system survey data you have collected, without consent from the participants.

Asking for help and guidance: if uncertain, have a discussion with your lecturers, your academic tutor, or Study Advice. 

Be aware:  

• Fraudulent representation of AI-generated work as your own constitutes plagiarism! 

• Reporting fabricated data constitutes fraud and is also a form of academic misconduct. Even if these data were produced by ChatGPT, responsibility and accountability lies with the human user! Therefore, it is important to thoroughly research a subject independently, and not trust the information provided in generative AI outputs, even if it appears to be cited. 

• When do I need to cite GenAI?

If you are using content that was created by a GenAI tool, e.g. a quote or an image, or referring to an interaction you had with it, then you will need to cite the GenAI tool as your source. Examples of how this citation should be formatted can be found at the relevant section of the UoR Citing References guide and at the Cite Them Right platform (see links below).

• When do I need to provide an acknowledgement of GenAI use?

In case you decide to use generative AI assistance (if permitted), it is important to acknowledge and accurately describe how this assistance was used in the production of your work.

An example of acknowledgment statement may look like this: 

"I acknowledge the use of ChatGPT 3.5 (OpenAI, https://chat.openai.com/) to produce a draft outline for further research. I confirm that no output generated by GenAI has been presented as my own work."

Check your assessment brief for a template acknowledgement statement and follow the instructions given there!

You may be asked to provide the following information to acknowledge use of AI:  

• Tool and version 

• Purpose of use 

• Stage(s) of the process in which it was used 

• How it was used: which prompts did you try and which outputs did you end up using? 

• What content was generated 

• How the generated content was evaluated and edited 

It is always advisable to keep a record of any interaction you have with a Generative AI tool, so you can return to it later and provide it as evidence in your work (for example, this information may be needed in a methods section, in an authorship statement, in an appendix, or as a footnote).  

Free products frequently come with a hidden price. A common implicit exchange includes giving companies access to user data. 

Before you subscribe to any service, check the agreement, and in particular their data privacy policy.  

Questions to ask yourself:  

• How are user data used by this company? 

• Are they shared and with whom? Only internally or with third parties? 

• Are they used for further training the AI model?

• Can I opt out from training these models? Do I want to? 

• Given the above, what information about myself do I feel safe sharing in my interactions with the AI agent?