AI Ethics Guide

Discussions about the ethics of AI use are evolving and they are nuanced by the discipline, experiences, and cultural values of staff and students, often leading to varied positions. In the spirit of transparency and effective use, as set out in the CUG ‘Position on AI Statement’, it is important to stress these evolving positions and the likelihood that students will need to continue to question their position on AI use as they enter the workforce.

To facilitate understanding and engagement, these conversations can be incorporated into wider issues of academic, professional, and personal integrity, assessment literacy, and the nature of research and authorship in the discipline. The sample tasks and discussion points set out below are intended to engage students in thinking about these wider ethical issues and how AI tool use might be one facet of consideration within these debates.

The sample tasks offer opportunities for students to explore the nature of academic integrity and authorship including reflections on their research and writing processes and on how AI tools might usefully be incorporated into these processes whilst mitigating ethical considerations. The sample tasks will benefit from adaptation to the discipline and have been designed to fit a variety of available time frames.

When discussing ethical issues and AI use, it can help to understand how AI tools work (e.g. Large Language Models or Grammar Checkers) and the role these functionalities have in processing and presenting data (Link to Guidance Page). The following section describes key ethical issues relating to AI tools (particularly GenAI) and includes further reading. Subsequent sections describe and resource activities for beginning conversations on AI and ethics.

Key Issues in Ethics and AI

Ownership and responsibility

The innovation of AI tools has reignited debates about the nature of authorship. This has implications in terms of where the liability for AI authored works lies and how to cite AI authored work.

Current Library guidance locates the publisher or owner of the AI tool as the author in citation:

A-Z Referencing Examples – Referencing in APA Style – LibGuides at Coventry University

How to cite ChatGPT

For further information, see:

Lee JY. (2023) Can an artificial intelligence chatbot be the author of a scholarly article?. J Educ Eval Health Prof, 20(6). doi: 10.3352/jeehp.2023.20.6

Stokel-Walker, C. (2023, January 18). ChatGPT listed as author on research papers: Many scientists disapprove. Nature. 613(7945), 620–621. https://www.nature.com/articles/d41586-023-00107-z[

Data integrity, corruption, and reproducibility

In addition to amplifying biases and misinformation, AI tools have been shown to degrade knowledge and language. This is a particular issue where newer generations of AI tools are unable to recognise the impact of earlier generations of tools or to distinguish between ‘human’ writing and AI writing, thereby exacerbating the problem.

Overreliance on AI in collating and summarising data may result in the production of misinformation, a reduction in the accuracy and veracity of knowledge, and a loss of trust in academic writing and scientific processes.

For further information, see:

Ball, P. (2023, December 5). Is AI leading to a reproducibility crisis in science? Nature, 624(7990), 624–627. https://www.nature.com/articles/d41586-023-03817-6

Fischer, T. (2024, September 13). How scientists debunked one of conservation’s most influential statistics. The Guardian. https://www.theguardian.com/environment/2024/sep/13/indigenous-factoid-nature-80-percent-false-biodiversity-aoe

Citations

Whilst Gen AI tools are increasingly able to reproduce citations, these remain inconsistent and unreliable. Even in the use of citing tools and embedded citations produced on publisher websites, citations should be manually checked for accuracy and consistency of style.

For further information, see:

Athaluri, S., Manthena, S., Kesapragada, V., & others. (2023, April 11). Exploring the boundaries of reality: Investigating the phenomenon of artificial intelligence hallucination in scientific writing through ChatGPT references. Cureus, 15(4), Article e37432. https://doi.org/10.7759/cureus.37432[1](https://apastyle.apa.org/style-grammar-guidelines/references/examples/journal-article-references)

Data Security

The opaqueness of some AI tool publishers and AI tools makes it difficult to be certain what the AI tool will do with any data inputted into it. This could have implications for data protection and GDPR, accusations of plagiarism and collusion, and the right to privacy.

LLM AI tools do not add inputted material to the existing data bank. However, some tools store prompts and inputs for the development of future AI tools.

For further information see:

Open AI. (April 2025). How your data is used to improve model performance. https://openai.com/en-GB/policies/how-your-data-is-used-to-improve-model-performance/

Employability

AI tools are transforming the workplace. People with traditional white collar, administrative, and creative roles are particularly vulnerable to automation. AI tools are now increasingly used in the application process and this has prevented candidates from reaching interview where their application does not meet the strict parameters of the AI tool.

For further information, see:

West, D. (May 2015). The future of work: Robots, AI, and automation. Brookings Institution Press. Available on locate: The Future of Work: Robots, AI, and Automation – Lanchester Library

The Environment

The data centres required to operate AI tools use much more energy than traditional search engines. These data centres also consume large quantities of water for cooling and minerals in their manufacture.

For further information, see:

Zewe, A. (January, 2025). Explained: Generative AI’s environmental impact. MIT News. Explained: Generative AI’s environmental impact | MIT News | Massachusetts Institute of Technology

de Vries, A. Oct 2023). The growing energy footprint of artificial intelligence. Joule, 7(10), 2191-2194. https://www.cell.com/joule/fulltext/S2542-4351(23)00365-3

Societal Impacts

Some AI tools have been fine tuned by workers on poorly paid contracts largely in the Global South. These workers ‘filter’ large quantities of upsetting and violent language and imagery.
Overreliance on AI tools has been identified as a possible cause of loneliness and poor rates of criticality and emotional intelligence.

For further information, see:

Zhui L, Yhap N, Liping L, Zhengjie W, Zhonghao X, Xiaoshu Y, Hong C, Xuexiu L, Wei R. (2024). Impact of large language models on medical education and teaching adaptations,
JMIR Med Inform, 12. doi: 10.2196/55933

Fatima, A., Shafique, M. A., Alam, K., Fadlalla Ahmed, T. K., & Mustafa, M. S. (2024). ChatGPT in medicine: A cross-disciplinary systematic review of ChatGPT’s (artificial intelligence) role in research, clinical practice, education, and patient interaction. Medicine, 103(32), e39250. https://doi.org/10.1097/MD.0000000000039250

Integrity issues pertaining to Academic Practice

Ownership and responsibility

Data integrity, corruption, and reproducibility

Citations

Academic Writing

In addition to prioritising Anglophone sources, LLMs tend to prioritise American standards of English. This can exclude contributors and participants who speak English outside of this context, e.g. international students, and impact on the language and syntax of academic English. It should also be noted that spellcheckers will tend to revert British, historical, and disciplinary or intentionally specific spellings to American standards.

For further information, see:

Louro, C. (2025). AI systems are built on English – but not the kind most of the world speaks. The University of Western Australia. AI systems are built on English – but not the kind most of the world speaks