In the context of Generative Artificial Intelligence (AI) tools - such as ChatGPT - it is important for both staff and students to explore the tools through the lens of their disciplinary area of expertise and practice in order to better comprehend the benefits and limitations the tools afford.

Among the disciplinary areas of validity, reliability and appropriateness, additional broader concerns that warrant consideration in the context of Academic Integrity in Higher Education include: 

  • Ethics using generative AI without appropriate declaration/referencing must be considered in terms of appropriate referencing of work that is not one's own. 

    For reference, at the time of writing, MIC’s
    Academic Integrity Policy considers that, among other factors, Academic dishonesty includes:

    ‘falsely representing the work of others as one’s own in an assignment.’ and ‘using co-authoring assistance in individual academic work, including the commissioning or purchasing of essay writing services, i.e. syndication.’ 

    However, detection in this fast-evolving space is very hard to prove and unreliable, even with the aid of detection tools such as Turnitin.

    Further ethical implications of the inappropriate use of Generative AI tools include the well documented fact that is that the content may not always be reliable - something termed by AI technologists as 'hallucinations' [1]. Such inaccuracies do not appear to be a critical concern to AI and BigData companies [2]. Among other factors that warrant concern are that, at present, Generative AI is trained on already dated material [as, for example, with GPT-4 up to Sep 2021; 3]  and may contain data that trains the Generative process or which perpetuate data-bound biases [4].

    As a result, student’s will need to critically review any content generated from Generative AI tools, form their own opinions and validate these through appropriate referencing.
     
  • Integrity of Assessment – There is a risk in using Generative AI to complete assessment tasks without an appropriate understanding of the assessment topic or critiquing of the materials that the learning outcomes of the module may not be met.

    Traditional written assignments are prone to this risk, though this can be mitigated through a considered review of the methods and approaches used in assessments.
     
  • Data Protection or Breaches – There are privacy concerns in using Generative AI tools from the perspective of data privacy and protection [5]. All users - staff, students and others - need to have regard for the appropriacy and potential later use of details submitted to AI tools.

    Caution should be applied as we are currently not able to trace the information and track further usage. See MIC’s
    Data Protection Policy 

Last Revised

April 2023.

References

Webb, M. (2023), 'Considerations on wording when creating advice or policy on AI use', National Centre for AI - Jisc, 14 February. Available at: https://nationalcentreforai.jiscinvolve.org/wp/2023/02/14/considerations-on-wording-ai-advice/ (Accessed on 12 April 2023)

Footnotes

[1] Wikipedia. Hallucination (artificial intelligence). (Accessed 18 Apr 2023)
[2] Heikkilä, M. Why you shouldn’t trust AI search engines. MIT Technology Review, Feb 2023. (Accessed 17 Apr 2023)
[3] OpenAI. Models: GPT-4. (Accesed 18 Feb 2023)
[4] Lund, B.D. & Wang, T. (2023), Chatting about ChatGPT: how may AI and GPT impact academia and libraries? Library Hi Tech News. (Accessed 18 Feb 2023).
[5] Almond, S. Executive Director, Regulatory Risk, Information Compliance Office (UK). Generative AI: eight questions that developers and users need to ask. Apr 2023. (Accessed 18 Apr 2023)