ChatGPT, AI and UAL Awarding Body qualifications
ChatGPT has been in the news a great deal lately. Centres have asked us at UAL Awarding Body for advice on how to manage potential student use of this software. This statement offers some initial guidance in a rapidly changing field.
ChatGPT is a state-of-the-art Artificial Intelligence (AI) language model developed by OpenAI that can generate human-like responses to various prompts. In the context of student work this may become an issue in terms of authenticating that a student has produced work themselves. We at UAL Awarding Body are clear that our qualification and assessment design and existing expectations of authentication already mitigate this new potential risk.
While UAL Awarding Body believes that the use of AI can be beneficial as part of the learning and development process, students are not permitted to submit work for summative assessment that has been generated by a chatbot or AI tool. This includes all formats in which such work could be produced, including text, imagery, video, sound, animation or any other outputs. Such work will be considered as malpractice, as the words generated are not the students' own. The only exception to this, is where the course content permits the use of AI generated work.
A well-rounded authentication approach can assist centres to detect occurrences where students have used AI/ chatbots to generate assessment evidence.
To mitigate risks centres should:
- Ensure they have sufficient supervision of students during all stages of the assessment process.
- Schedule as much of the assessment time within normal classroom hours where they can have sufficient oversight.
- Take steps (if necessary) to limit, block or ban AI/ Chatbot websites during the assessment period.
- Remind students that the use of AI / Chatbots will constitute cheating and the consequences of doing this.
- Hold regular tutorials so they are familiar with the students' work.
- Ensure that should a student submit AI / chatbot generated work, this is appropriately referenced. This will ensure that should plagiarism detection tools be used, that the work will not be flagged as malpractice.
Evidence of developmental work, use of research and reflection is crucial in the authentication process to ensure that the work submitted is original and has been completed by the student themselves. This must include keeping records of the thought process, drafts, and revisions made during the course of the project, along with any relevant documentation or software tools used. Regular tutorials will ensure teachers are familiar with the usual working style of their students.
Whether the student has included evidence of developmental work: ensuring that students maintain records of their thought process, drafts, and revisions made during the course of the project can help ensure that the work submitted is original and has been completed by the student themselves.
Use of plagiarism detection software: AI-powered plagiarism detection tools can quickly and accurately identify instances of plagiarised text or ideas, but they may not be able to detect creative plagiarism such as paraphrased content or original work that has been heavily influenced by existing material.
Writing style evaluation: While AI plagiarism detection algorithms can be trained to identify the writing style and tone of a particular student, it is possible for students to manipulate their writing style to avoid detection. Therefore, tutors need to ensure that they have in place sufficient oversight of students work and abilities so that they are more likely to be able to detect irregularities in their writing style.
Real-time support: AI bots can be beneficial to provide real-time support to students during the course of the project, but it is important to monitor the use of these tools to prevent students from copying and pasting work from online sources.
Limitations of AI: AI algorithms may not be able to accurately evaluate creative works such as artwork, poetry, or music, and may not be able to identify instances of plagiarism if the source material is significantly different from the original content.
To provide a robust approach to the authentication process for student work, colleges need to implement a multi-faceted approach that takes into account the limitations of AI technology and utilises traditional methods in addition to AI-powered tools. For example, incorporating peer review, using multiple plagiarism detection tools, and requiring students to present their work in person can help ensure the authenticity of student work. By taking these steps, colleges can improve the authentication process and maintain the credibility of their academic programs.
While the capabilities of AI platforms such as ChatGPT are currently limited, UAL Awarding Body acknowledges there will be rapid developments in this area as AI continues to evolve. The capabilities of AI and impact of AI-human interaction have huge implications socially, economically and politically. These developments are at the cutting edge of learning and creativity, with potentially very enabling and exciting applications. We will review the landscape regularly to ensure emerging risks are mitigated and ensure our students continue to benefit from the advantages AI has to offer for their future learning and employability.