Guidance for students on the use of Generative Artificial Intelligence

AI tools draw on a vast range of data typically 'scraped' from the internet. Generative AI is a form of AI that uses this data to 'train' a tool that can then generate content. These tools allow users to generate content by simply entering a plain-text or speech input, known as a 'prompt.' Content can include text, images, speech, music or computer code.

Most genAI tools use machine learning to extract meaning from text or speech or predict something based on available data. For example, sophisticated algorithms are used alongside Large Language Models (LLM) in ChatGPT to determine the most likely word to follow on from the last. In a crude way, it is paraphrasing what is already written and openly available on the Internet. It is a little like existing spellcheckers, grammar checkers or predictive text functions, but far more sophisticated. The data it draws on is far greater in scale.

GenAI creates brand new, completely original variations of content. It can also modify, summarise and potentially improve existing content. These tools can tailor outputs to the tone, topic and audience requested by the user. While not simply copied, these new outputs preserve a likeness to original data. Essentially, they mimic existing content.

It is equally important to understand what genAI is not, and what it does not do. It does not currently automate tasks without significant adaptation. Nor does it replace people or jobs. It will, however, affect the way we all work and the jobs we will do in the future.

It is important that you understand the limitations of any AI system. GenAI is not a trusted expert. However, it often speaks in an authoritative style which implies a false level of confidence and accuracy. Always check the factual accuracy of any content that is generated for you.

GenAI outputs are often superficial and simplistic. They can be stereotypical or prejudicial. Generative AI LLMs draw on a wide variety of sources without checking for accuracy or appropriateness. They have been found to replicate and perpetuate biases including racism, sexism, ableism, homophobia and transphobia, and other prejudicial or inappropriate content. They may align with particular political or commercial agendas. Do not rely on AI generated content without question.  

GenAI mimics human outputs but does not 'understand' or 'know' anything itself.

Over-reliance on AI tools to generate written content, images, software code or analysis reduces your opportunity to practise and develop key academic skills (for example, writing, critical thinking, evaluation, analysis or coding skills). These are all important skills that are required to succeed in and beyond your time at university.

Academic integrity involves a commitment to the core values of honesty, trust, fairness, respect and responsibility in all academic endeavours. While genAI tools may enhance your learning and preparation of assessments, they are not a replacement for your own effort and understanding. They must not be used as a means of gaining unfair academic advantage. GenAI is a tool not a source.

Section 24 of York St John's Code of Practice for Assessment has been updated to clarify unacceptable uses of genAI. For example, it makes clear that use of genAI that is not properly acknowledged may constitute plagiarism (24.3.1 p110). All work submitted for assessment should be your own original work.

Tools to detect the use of geAI, including that embedded in Turnitin, are currently prone to return significant false positives and false negatives. As a result, it is not appropriate for York St John to deploy these tools at this time. This will be kept under frequent review as the technologies develop.

You should explicitly acknowledge where, and how, you have used genAI tools. When you submit a written assessment, you will be required to complete a pro forma declaration to this effect (for example, on a cover sheet or equivalent).

It is not appropriate to quote from a genAI output as if it were an authoritative academic source. If you do choose to quote from a genAI output, for example, when discussing AI in an assessment, you must provide full and accurate citations. Emerging citation protocols differ by citation style. Please follow the York St John referencing guidance. Please check regularly for updates.

Use of text composition tools and services, such as spell checkers, thesaurus or proofreading tools, may be powered by more rudimentary AI. These are acceptable to use and do not require citation if they only influence the form but not the content of an assessment or other output.

It is not enough to paraphrase or make minor modifications to genAI outputs. It is essential that you actively use academic skills, such as analysis and critical evaluation. Doing so will enable you to meet your programme learning outcomes and assessment criteria, and therefore succeed on your course.

AI tools can generate 'hallucinations' including inventing non-existent sources, using offensive language, or making inaccurate or defamatory claims about individuals or organisations.

Given the issues of accuracy, bias and hallucinations outlined above, it is the responsibility of each student to evaluate critically the appropriateness and accuracy of AI generated content.

As with many other technologies, genAI tools have environmental and socioeconomic impacts. Some report ethical concerns about the carbon emissions generated by training processes for LLMs. Others highlight the use of cheap outsourced labour in the development of the tools. It also raises long term questions about the automation of jobs and potential exploitation of workers. It has the potential to exacerbate global structural inequalities. Students may wish to make personal choices on how and whether to use genAI tools on this basis.

Do not submit sensitive, proprietary or confidential information to genAI tools. Any information entered into a genAI tool can become part of its LLM training set. This could result in such information being used in outputs. The owner of the tool may also share this information with third parties without your permission.

Staff and students must comply at all times with applicable copyright law and appropriate consents and licences. It is inappropriate and possibly illegal to enter into genAI tool any copyright, proprietary or confidential information. This includes your set readings, resources from the library (physical or online), or copyrighted content from the internet. All material recorded as part of a lecture recording is subject to copyright legislation. You may wish to enter your own lecture notes into a genAI tool, but must not enter a lecture transcript or extract from a transcript.

Likewise, it is not acceptable to enter into genAI tools any personal identifiable information (PII) about yourself or others. This could be used by hackers to compromise cyber security. PII might include names, addresses, family details, sex, gender, race, disability, passwords, reference numbers (student number, driver's licence, passport, visa, financial information), social media accounts, medical records, significant dates, details of a student's course and so on.

Any current or future incorporation of genAI into teaching or assessment practice at York St John will adhere to high standards of inclusive practice articulated in our Inclusive Education Framework.

While some genAI tools are free to use, many reserve their strongest functionality for paid subscriptions. This has the potential to limit equity of access, affordability and support. Where York St John teaching or assessment require the use of genAI tools, we will give full consideration to financial barriers and unexpected course costs.

While we will assume that students have some access to and make some use of genAI, in line with sector guidance, we will seek to ensure 'as far as possible, that no advantage can be accrued by students who can access more advanced Generative Artificial Intelligence tools from behind a paywall' (QAA 2023, PDF, 1.12 MB). Therefore we will only assume access to and use of free versions of genAI tools.

The long-term societal changes brought about by AI technologies are not yet known. They are likely to be far reaching. We will learn together. University education plays a key role in preparing students for their future. The University will work in partnership with students and staff to:

• Ensure AI literacy and wider digital literacy are embedded in curricula as appropriate.
• Review and enhance practices of learning, teaching and assessment to ensure they remain relevant, rigorous, authentic and meaningful.
• Equip staff with the training and understanding of genAI that will allow them to support students’ success.
• Gather and act upon feedback from students, staff and employers regarding our use of genAI in learning, teaching and assessment and in research.
• Establish a generative AI community of practice that enables academic and related staff to share insights and best practice in the use of genAI in learning, teaching assessment and student support.

AI is an emerging and rapidly changing technology presenting many opportunities. It is a disruptor in higher education, as in many other sectors. We will ensure we support staff and students to develop their AI literacy and identify the affordances it may provide for learning, teaching and assessment.

The practical and ethical considerations of the use of AI will develop over time. Its capabilities and limits will change. We will review our AI guidance regularly to ensure it is current, practicable and appropriate to the ethos and approach of York St John.