Golden Rules of AI Use

Upholding Academic Integrity: The Golden Rules of AI Use

What is Academic Integrity in the AI Era?

Academic integrity forms the bedrock of higher education, representing a commitment to honesty, ethical conduct, and intellectual rigour. Fundamentally, it means being truthful in your academic pursuits, acknowledging all sources of information, and ensuring that the work you submit genuinely reflects your own understanding and effort. This commitment necessitates meticulous referencing, clear citation practices, and a steadfast avoidance of plagiarism, which includes not allowing others to use your work or submitting work that is not your own.

In the context of AI, many UK universities have refined their policies to address the concept of "false authorship". This term refers to the creation or modification of academic work, either wholly or in part, with unauthorised or undisclosed assistance from individuals (such as essay mills or proofreaders operating beyond permissible scope) or technology (including generative AI and machine translation tools). The critical aspect is the extent to which this external help diminishes the student's genuine intellectual contribution, making it so substantial that the student can no longer be considered the true author of the submitted work. The distinction is crucial: if the use of AI reduces a student's active involvement and understanding, replacing their personal authorship and critical input, it crosses the line into academic misconduct.

Prohibited Uses: Clear Examples of Academic Misconduct

Universities across the UK have established clear boundaries for AI use to maintain academic standards. Certain applications of AI are consistently identified as forms of academic misconduct:

• Submitting Unedited AI-Generated Content: This is universally regarded as a severe offense. Presenting academic work that is partially or entirely generated by AI without significant personal review, understanding, or original modification constitutes false authorship. The University of Portsmouth's "Golden Rule" explicitly states that merely editing AI-generated content is insufficient for it to be considered a student's own original work, highlighting that superficial changes do not equate to genuine intellectual effort.

• Over-Reliance on AI: Excessive dependence on AI tools that impedes a student's learning, misrepresents their true academic capabilities, or substitutes for active engagement with the course material is prohibited. This includes using AI to regularly assist with mathematical reasoning, coding, or complex analysis, which can undermine a student's ability to develop these core skills independently.

• Uncritical Acceptance of AI Outputs: Students are expected to exercise critical judgment. Accepting AI suggestions without thorough evaluation, careful review, or consideration of their accuracy, context, or appropriateness is considered misconduct. Students are ultimately held accountable for any errors or biases present in AI-generated content they incorporate into their work.

• Using AI to "Prepare" Your Work: University policies explicitly forbid having any other person, service, or AI tool "prepare your work" through copying, translating, or lightly editing existing material. This prohibition extends to using AI translators to convert assessments into English before submission, as machine translation can be treated as false authorship if it replaces the student's own linguistic and conceptual effort.

• Presenting AI Outputs as Your Own Original Work Without Acknowledgment: Any inclusion of AI-generated text, images, audio, video, mathematical formulae, or computer code in an assessment without proper and transparent acknowledgment constitutes a direct violation of academic integrity.

• Citing Unverified AI-Found Sources: Relying on AI to generate references without personally checking their existence, accuracy, and relevance (often referred to as "phantom citations") is a breach of academic integrity. Students are required to only include references to information they have genuinely accessed, read, and used themselves.

• Uploading Course Materials to AI Tools: Uploading confidential course materials, including assessment guidance, slides, readings, or transcripts, to any generative AI tool is prohibited due to potential violations of copyright law and data privacy.

• Using AI-Generated Materials as Research Sources: Generally, AI-generated material should not be used as a primary research source in submitted work, unless the assessment specifically requires a critical analysis or commentary on the AI system's outputs or functionality itself.

The consistent emphasis on "false authorship" across multiple university guidelines signifies a crucial evolution in the definition of academic misconduct. It expands beyond the traditional understanding of plagiarism (direct copying without attribution) to encompass any unauthorized external assistance – whether human or technological – that fundamentally compromises the student's intellectual ownership of the work. This indicates a shift in focus from merely detecting copied text to evaluating the degree of intellectual input and understanding demonstrated by the student, irrespective of the tools employed. The "Golden Rule" that "editing AI-generated content is not enough" further reinforces this, making it clear that superficial modifications do not equate to genuine authorship.

The Risks of Detection and Consequences

While detecting AI misuse can be challenging, universities are actively developing and implementing strategies to identify it. Should suspicion arise, students may be requested to provide draft work, research notes, or attend a panel hearing to explain their work production process and demonstrate their understanding. This mechanism allows institutions to verify the authenticity of the submitted work and the student's intellectual engagement.

Academic misconduct, including false authorship resulting from AI misuse, is treated with utmost seriousness across UK universities. Such violations can lead to severe implications for a student's grades, academic standing, and overall studies, potentially resulting in failing grades, suspension, or even expulsion.

The juxtaposition of high rates of undetected AI-generated content (94% in some studies ) with the severe consequences for academic misconduct creates a significant ethical tension for both students and institutions. For students, this scenario can present a high-risk, high-reward temptation, where the perceived low probability of detection might encourage misuse despite the severe penalties if caught. For institutions, this highlights the inadequacy of punitive measures alone and necessitates a greater emphasis on fostering a culture of integrity, transparency, and education regarding responsible AI use, rather than solely relying on detection technologies that are demonstrably unreliable and prone to false positives. This situation prompts a shift towards proactive ethical guidance and assessment design that inherently mitigates AI misuse, rather than relying solely on reactive detection.