How to Detect AI Writing in a Student A-Level EPQ Safely

The Extended Project Qualification (EPQ) is widely regarded as one of the most rigorous and independent components of the A-Level curriculum in the UK. Designed to bridge the gap between secondary education and university-level study, it requires students to conduct extensive primary and secondary research, maintain a detailed production log, and ultimately produce a comprehensive 5,000-word essay or a related practical artefact. However, because the vast majority of this work is completed independently outside of the classroom over several months, the A-Level EPQ has become particularly vulnerable to the undeclared use of generative artificial intelligence tools like ChatGPT and Claude.

For UK secondary school teachers and EPQ supervisors, this presents an unprecedented challenge. Ensuring academic integrity is paramount, but accusing a student of using AI to write their A-Level EPQ is a serious allegation that can severely impact their academic trajectory and confidence. The traditional methods of spotting plagiarism are no longer sufficient, and educators must adapt to a landscape where AI assistance is readily available. In this post, we will explore why the EPQ is so susceptible to AI generation, how to evaluate likelihood scores fairly, and why your professional judgment remains the ultimate safeguard.

The Challenges of Spotting AI in an A-Level EPQ

Spotting AI-generated content in a massive document like an A-Level EPQ is entirely different from catching a copied-and-pasted paragraph from Wikipedia in a standard homework assignment. Generative AI models do not simply copy existing text; they predict the next most likely word based on vast training datasets, creating entirely novel sentences that will completely bypass traditional plagiarism checkers. This means that a student could theoretically generate an entire EPQ literature review that scores a perfect 0% on standard similarity software.

Why Traditional Plagiarism Checkers Fail

Traditional plagiarism detection software works through exact string matching and structural similarity comparisons against an established database of published works, past essays, and internet pages. When a student uses Claude or ChatGPT to write their A-Level EPQ, the resulting text is technically original. The AI effectively synthesises the information, meaning traditional checkers are searching for a direct source that simply does not exist. Relying on these outdated tools gives teachers a false sense of security and leaves the EPQ assessment process deeply flawed.

The Risk of False Positives

The converse problem is just as dangerous: the risk of false positives. Many teachers have turned to early AI detection tools that operate purely by looking for 'burstiness' (variation in sentence length) and 'perplexity' (how predictable the vocabulary is). Unfortunately, these metrics often penalise students who write in a highly structured, academic, and formulaic manner—exactly the kind of writing style that the A-Level EPQ encourages. Falsely accusing a high-achieving student of using AI because they wrote a clear, methodical essay can destroy the fragile teacher-student trust and demotivate the student entirely.

Using Likelihood Scores and Probabilities

To navigate this complex landscape, the educational technology industry is shifting away from binary "AI vs. Human" judgments and moving towards a probabilistic approach. When reviewing an A-Level EPQ, teachers should not be looking for absolute certainty, but rather an indication of the probability that AI was heavily involved in the drafting process.

Moving Away From Absolute Certainty

No software in the world can currently guarantee with 100% accuracy whether a specific paragraph was written by a human or a machine. Generative models are constantly evolving, and human writing styles are incredibly diverse. Therefore, when attempting to detect AI writing in an A-Level EPQ, teachers must accept that there is a margin of error. Shifting the mindset from "proof of cheating" to "indication for further investigation" is the most crucial step a school department can take when updating their academic integrity policies.

Understanding a Probabilistic Scale

Instead of relying on a definitive red flag, modern educational platforms use a probabilistic scale. GradeOrbit's dedicated AI detection capabilities analyse the text and return a likelihood score ranging from 0 to 100%. A score of 85% does not mean that 85% of the A-Level EPQ was written by AI; rather, it indicates an 85% probability that the text exhibits characteristics strongly associated with generative language models. Teachers can use this likelihood score as an initial smoke alarm. If the score is low, the teacher can proceed with their standard marking. If the score is exceptionally high, it triggers the next, most critical phase of the process: human evaluation.

Combining AI Detection with Professional Judgment

Software should never be the sole arbiter of a student's academic fate. A high likelihood score on a section of an A-Level EPQ must always be paired with the teacher's professional judgment. You know your students, you understand their baseline capabilities, and you have watched their projects evolve over the academic year. These insights are invaluable and completely inaccessible to any AI software.

Cross-Referencing the EPQ Logbook

The A-Level EPQ contains a built-in safety mechanism: the production log. This logbook is supposed to comprehensively document the student's journey, detailing their planning, research struggles, shifts in project focus, and drafting timelines. If a student submits a flawless 5,000-word essay with a high AI likelihood score, but their production log is sparse, generic, or lacks specific timeline details, this provides strong circumstantial evidence of AI generation. Conversely, if a student has an extensively detailed logbook that perfectly aligns with a highly sophisticated final essay, a high AI score might merely indicate the use of AI for proofreading or restructuring, which you can discuss with the student.

Recognising the Student's Authentic Voice

Throughout their secondary education, students develop a distinct academic voice. While the A-Level EPQ expects a step up in academic terminology, drastic and sudden shifts in tone, vocabulary, and syntactic complexity are major red flags. If a student who typically struggles with basic comma splices is suddenly producing perfectly balanced, multisyllabic arguments about complex socioeconomic theories, your professional judgment should immediately be engaged. You can cross-reference the EPQ with handwritten mock exams or previous controlled assessments to compare the genuine voice with the submitted text.

Having the Conversation with Students

If the AI likelihood score is critically high and your professional judgment tells you that the student's A-Level EPQ is not entirely their own work, you must approach the conversation with care. Inquisitorial tactics often backfire, leading to defensive denials. Instead, use an investigative, supportive approach.

Constructive Approaches to Suspected AI Usage

Rather than demanding a confession, ask the student to explain specific, complex concepts from their EPQ. Ask them to justify why they chose a particular source over another, or how they synthesised conflicting viewpoints in their conclusion. If a student has genuinely researched and written the piece, they will be able to discuss the topic fluently, even if they stumble over their words. If they used ChatGPT to generate the entire essay, they will quickly find themselves unable to explain the nuances of their own submitted work. Furthermore, you can use this as a teachable moment to explain the boundaries between using AI as an assistive brainstorming tool and using it to outsource the cognitive heavy lifting of the A-Level EPQ.

Conclusion

Managing the integrity of the A-Level EPQ in the age of generative AI is undeniably difficult, but it is not impossible. By understanding the limitations of traditional plagiarism checkers, utilising sophisticated probabilistic likelihood scores, and firmly relying on your own professional judgment and knowledge of the student, you can ensure that the assessment remains fair and robust. The goal is not to police every keystroke, but to maintain the EPQ's status as a genuine reflection of a student's independent academic capability.

Try GradeOrbit for Honest EPQ Assessment

If you are supervising an A-Level EPQ cohort and struggling to balance the demands of marking with the need for academic integrity, GradeOrbit is here to help. Built specifically for UK secondary school teachers, GradeOrbit provides a robust, probabilistic likelihood score from 0-100% allowing you to assess student work with confidence. Our system allows you to choose between faster, 1-credit AI models and smarter, 3-credit models depending on the depth of analysis you require. Don't let AI anxiety slow down your workflow. Sign up for GradeOrbit today and take the guesswork out of EPQ supervision.