Following a meeting of the General Board’s Education Committee (GBEC) on Wednesday 24th January 2024, an amendment was made to the existing academic misconduct statement regarding the use of Large Language Models (LLMS) such as ChatGPT and Microsoft Copilot, often referred to as generative artificial intelligence (AI). The revised statement now reads:
“A student using any unacknowledged content generated by artificial intelligence within a summative assessment as though it is their own work constitutes academic misconduct, unless explicitly stated otherwise in the assessment brief.” Plagiarism & Academic Misconduct
This revised statement aims to provide more specific clarity around the use of these tools in summative assessments whilst allowing for discipline-specific definitions of what is appropriate use to be made. The change enables staff and students to engage with these tools more in independent study and formative work, encouraging open dialogue about suitable use, ethical implications, the consequences of over-reliance, what stands to be gained, and what could be lost in the use of these emerging tools.
We recognise that answers to these problems are as diverse and varied as the myriad questions being asked about LLMs, their potential impact on education and the world at large, so this guidance and the updated statement are designed to support local discussions within departments and faculties and facilitate the development of discipline-specific statements on the use of LLMs. These statements should primarily address use within summative assessment and serve as guidance for formative and independent practice whilst recognising the variety of individual approaches where appropriate.
We recommend using this guidance together with the information available in the Suspected Academic Misconduct – Staff Guidance Document from The Office for Student Conduct, Complaints and Appeals (OSCCA). Large Language Models require the use of existing information to derive their responses so anything generated has been informed by the work of others without appropriate attribution.
Discussing Large Language Models and Generative Tools
It's understandable that academics, both in their role as teachers and as examiners, might have concerns about the legitimate use and misuse of LMMs in academic work. Here are some suggestions on how a Faculty or Department might address these concerns:
- Clear Guidelines: Provide clear guidelines for students and those teaching, on the use of LLMs in both assessed and non-assessed work. Specify what is allowed, what isn’t, and why. For example, using LLMs for research and generating ideas might be acceptable, but directly copying generated content may not be because gathering information might not be the focus of the assignment but independent synthesis and explanation is.
- Promote Process over Product: Encourage students to use LLMs as a supplement to their own work, not a replacement. Students could focus more on the processes involved in problem solving, project iteration or essay writing over the final product. The goal should be to enhance their learning, not bypass it. The inherent benefits and focus of a programme of study or discipline should be shared with students and promoted regularly.
- Educate Students: Make sure students understand what constitutes plagiarism and the consequences of engaging in it and explain how LLMs could and shouldn't be used in academic work. Discuss the intended outcomes, for both the formal criteria and personal skill development, and how the student may inhibit their own progression through over-reliance on these tools. Inform students of the ethical implications involved in using these tools and how it may impact their work and process. These risks might include the misappropriation of information, lack of acknowledgement of sources, sharing of own work to unknown data repositories, environmental impact of utilising these tools.
- Inform Staff and Supervisors: Discuss the potential uses of LLMs by students and how they should be monitored and resolved by staff. Ensure that all staff involved with a programme of study, assignment, or teaching a cohort of students are aware of the approach being taken and can communicate this approach to students confidently or share up to date information with them.
- Open Dialogue: Maintain an open dialogue with students and those who teach them about their use of LLMs. Encourage students to discuss if and how they're using these tools to enhance their learning and where they might be risking over-reliance or potential academic misconduct. Support students and teachers in exploring the potential of these tools openly whilst recognising the possible risks.
- Assessment Design: Design assessments in a way that requires critical thinking, personal reflection, or specific knowledge that can't easily be replicated by LLMS and generative tools. This may also include opportunities to reflect on intended learning outcomes required of an assessment and identify where these tools could be utilised in support of the students’ work whilst not negatively impacting their learning and development.
LLMs and Generative Tools Rubric
|
LLM Tool Use |
Description |
|
Topic Selection |
The software can suggest essay topics based on a general theme or subject area. |
|
Research Assistance |
The software can collate and present information and data on a given topic. |
|
Summarising Literature |
The software can summarise literature, analyse information and provide insights |
|
Outline Creation |
The software can help create a structured outline for the essay, or part of an essay. |
|
Generating Text |
The software can assist in writing the essay including writing introductions, body paragraphs, and conclusions. |
|
Data Analysis |
The software can analyse, summarise, and visualise qualitative data, providing insights for further evaluation. |
|
Editing and Proofreading |
The software can suggest improvements in grammar, punctuation, sentence structure, and word choice. |
|
Citation and Referencing Guidance |
The software can provide guidance on how to cite sources and create a bibliography in a specified style. |
|
Paraphrasing |
The software can be used to paraphrase existing content to make it appear as original work. |
|
Translation |
The software can translate content from one language to another. |
|
Code Generation |
The software can write feasible code in a variety of languages to meet the prompted brief. |
|
Interpret Images |
The software can “read”, interpret, and communicate the contents of an image including text. |
|
Image Generation |
The software can generate images and diagrams to varying degrees of accuracy. |
|
Mathematical Problems |
The software can be used to explain solutions to mathematical problems and, in some cases, provide solutions. |
|
Informed Feedback |
The software can provide feedback on a piece of work, this can reference criteria and rubrics if provided. |
|
Inspiration & Direction Setting |
The software can suggest sentences or paragraphs which are used to inspire a student’s own line of thought. |
|
Discussion & Development |
The software can provide a critical “partner” for students to discuss ideas and ask questions to further their own understanding. |
Further Support & Guidance
More guidance on these areas and how to approach them will be developed and provided to departments and faculties to inform decision making. If you have any questions, or would like to discuss local approaches and concerns, please contact the Blended Learning Service (mailto:info@blendedlearning.cam.ac.uk).