Enterprise AI Analysis

Learners consistently drew a clear line between using AI to enhance learning and using it for assessed work. They viewed GenAI as a valuable tool for scaffolding thinking, structuring arguments, clarifying content, and low-stakes practice, but emphasized it must be kept separate from anything that is assessed. As one learner stated, 'Never in tests, not in the exams... but helping with homework and stuff' (FG3) is acceptable. Brainstorming and analysis assistance was considered legitimate intellectual support, provided the learner did the substantive writing themselves. Copy-pasting AI output was widely condemned as plagiarism.

The underlying principle articulated was authorship: 'If you didn't create it, it shouldn't count as your work' (FG6). Acceptable uses were framed as formative (drafting outlines, practicing recall, checking understanding), summarised as 'As long as it's not for grades, it's OK' (FG3). Any use that substituted for assessed performance was considered grounds for disqualification.

A second boundary for GenAI use was drawn based on subject area. In STEM contexts like maths, GenAI was seen as a useful tool for procedural guidance and instant feedback, helping with 'maths equations' (FG3). Conversely, in humanities (e.g., English), where authorship and expression are key, learners argued AI 'defeats the purpose' (FG4). The line was even sharper for expressive arts, where personal voice and aesthetic are paramount.

Learners generally supported teachers using GenAI for impersonal, low-stakes administrative tasks and resource creation. This included drafting lesson slides, generating extra practice material, and creating quizzes ('Making revision games, like Blookets or Kahoots, is fine.' - FG2). However, a common expectation was that teachers must fact-check any GenAI content before using it in class ('fine as long as they check it' - FG8).

Many students sympathized with teacher workload, accepting AI use to ease administrative burdens. The consensus was that machine-drafted templates were tolerable only if teachers 'made it theirs.' As one learner summarized, 'For huge admin things, maybe, but not for things that matter to us' (FG4).

Regarding feedback and reporting, opinions were mixed but leaned negative. Writing pupil reports and formative comments were seen as core teacher responsibilities requiring first-hand knowledge of the learner, making GenAI inadequate. 'It defeats the purpose of the report a bit because it's supposed to be individualised feedback, but the actual tool knows nothing about you' (FG5). Some viewed staff reliance on GenAI as undermining the teacher's role or 'lazy' (FG4). Others were okay with GenAI for basic report generation if personal commentary remained human or if it helped strip out bias. Crucially, many participants felt it was 'hypocritical' (FG8) for teachers to ban student AI use while using it themselves, raising questions about the school system's consistency.

A significant minority of students avoid GenAI altogether, citing concerns about reliability and accuracy ('it's wasted time using it, getting wrong answers isn't worth it' - FG1) or fearing academic consequences of undetected errors ('If I was to write down something wrong in Advanced Higher I fail, so I don't use that at all' - FG1). Others worried about developing an over-dependence that could erode future competence ('When you use these kind of stuff, it's like you're depending on them... that feeling of not being able to do any work is quite scary' - FG1). Some stated their risk-aversion bluntly: 'I haven't used it. I don't risk these kind of things' (FG1).

Rules on AI varied significantly across classes and teachers, leading to uncertainty and anxiety. This inconsistency ('if I use AI for business and [teacher 1] the head of department saw me, I would get [reprimanded]. So whereas [teacher 2] and [teacher 3], they wouldn't probably actually mind that much. So I don't think it is consistent around the school' - FG3) caused worry about accidental misconduct and calls for clearer guidance.

Students attributed inconsistent GenAI policies to uneven staff expertise, with many teachers defaulting to prohibition if they lacked understanding ('Some teachers don't even really know what ChatGPT is, so they just say 'don't use it'' - FG8). They called for compulsory professional development for teachers to model responsible practice and help students avoid plagiarism or misinformation. Learners wanted explicit instruction on how to interrogate AI output, check sources, and recognize bias, extending beyond mere prompting to critical evaluation and skepticism. They also raised concerns about AI's environmental cost.

Learners voiced anxieties about long-term dependence on GenAI and its potential to dilute personal style ('If you use AI for like creative purposes like music or art, you don't really develop your own style with that' - FG4) and weaken cognitive skills ('if you use AI... you can't do your working, so you're obviously not learning' - FG6). They feared relying too heavily on AI could lead to mistakes in professional contexts (e.g., medicine) and make it difficult to write by hand again after using AI for essays.