Earlier this year, our BILT Student Fellows (Pratham and Gaurav) ran a series of sessions on AI and Inclusion, one for each of the Faculties. We opened each one with a panel talk from the PVC Education, a senior member of Faculty staff, and myself. There followed small table discussions on a range of assessment-themed topics. I sat with the theme of coursework in assessment.
In this blog, I share some highlights from the fascinating discussions (and thanks to all the students who participated with so much passion in each session!). This echoes research undertaken by BILT teams and colleagues at the University that informs decision-making and sense-checking through the student voice. Most of the blog refers generically to AI, which most students took to mean generative AI unless otherwise specified.
You can also read Joe Gould’s blog about student reflections on using AI for reflective essays and Amy Palmer’s blog about student reflections on using AI for group work from the same event series.
What do students like about AI?
Lots of students noted how AI is great for answering questions quickly, especially when they have limited time with staff or feel shy to ask “stupid” questions. International students noted how AI helps them translate Western-centric lecturers’ norms and phrases. It is also used to supplement vague feedback provided by staff.
A repeated observation was how many students said the AI helps them clarify something their teacher said or something a teacher didn’t explain well. One student explained how everyone learns differently, and with AI, they can get it to keep articulating something in a myriad of different ways with AI until it works perfectly for them.
Several students noted that sometimes the essay question is so vague that everyone goes to AI to ask it to explain, that there is often a lack of understanding of what the teacher wants from students and AI fills that gap when it feels intimidating to ask. One student noted how staff have changed their practice to spend lots of more time to explain what they want from the essay question in dialogue with students, with examples; this is great practice, and the student laments it didn’t exist before!
AI is used for lots of tasks:
- Research searches and article summaries, and to highlight things students might otherwise overlook.
- Students use AI like a critical friend, asking it to challenge their writing. Sometimes the AI is helpful and other times not so much. One student noted how they lean on AI to have conversations when they don’t have a group of peers to do so, stating that it’s interactive and “feels natural”.
- The use it to sense-check what they are writing. Lots of students enjoy the sense of validation they get from the AI.
- For one neurodivergent student, they find that AI helps them take abstract examples discussed by their teacher and translate it into concrete examples to help them understand more clearly.
Academic misconduct and AI
Lots of students complained that their peers are flagged for AI misuse when they haven’t used it at all, and don’t know why they were flagged. At the same time, they observe that other peers use it and “get away with it”. For those flagged for misuse, they note the stress and fears are overwhelming, the punishment is huge, and it affects how students use AI. Again and again students reiterated the need for teachers to be very clear what is and isn’t allowed, especially when everyone interprets rules and policies different ways.
They also lacked clarity on the processes involved in flagged and investigating students who were flagged. They noted the risks around calibration with individual assessors taking different positions on AI use, asking what this means for fairness – students had diverse experience of this, and it caused them a lot of stress and uncertainty. One neurodivergent student was concerned with Turnitin’s high potential to label neurodivergent work as misuse of AI.
Several students stated they felt obliged to pay for a plagiarism checker service to review their coursework and ensure it isn’t flagged (even when they have 100% not used AI in the first place). This was most common in Health and Life Sciences, and Science and Engineering.
Lots of students worry about how using AI is stealing their voice, and worried they will never develop their own writing voice due to over-reliance on AI (more common concern in Arts & Social Sciences). One student who speaks multiple languages noted that their writing appears AI-like, and they have grown paranoid about their writing in case they are accused of misuse, including the use of known AI buzzwords.
What do students dislike about AI?
A repeated concern was how uncritical many students are about AI: “People do blindly trust what the AI says, and are surprised when it gets something wrong”. AI doesn’t provide a depth of knowledge, it is too superficial, according to several student observations. Students note that AI is quick, but it is not accurate and overlooks context. One student found that “use of AI means less critical thinking, so it should not be used for coursework. If you can’t do it, it hinders you in the real work, it’s what you are in university for”. They questioned the permission to use AI in coursework “what are you being assessed on”, and to mitigate this concern suggested more practical assessment tasks and personal reflection.
Some students also raised concerns about growing dependencies on AI, that it “might be dangerous if we rely on it too much, you have to develop confidence yourself”. Students noted that AI reliance is growing and it is important that students in early years of their degrees develop knowledge and skills through various processes to avoid learning loss. As one student put it: “If I use AI too much I don’t know how to do it myself, I can’t decide it it’s right or not. It doesn’t always match up with what the paper says, so I had to force myself to learn the skills to do it myself”. When over-relying on AI “you miss out on the skills you could have developed”. Sadly, as one pointed out “lots of people don’t care about learning”.
These discussions prompted students to reflect on why thy are taught what they are: “sometimes we don’t see why we learn something – but if we know why it was important, I wouldn’t just memorise it short term. I want to be able to do my job when I leave university, and not rely on AI. Sometimes it will take me longer to ask AI. I am paying £9K, why am I educating AI instead of myself?”.
Almost all students observed digital equity issues between those with premium licenses and without. They also noted that some students have higher AI literacy and therefore at an advantage compared to their peers. As one student put it, premium licenses “exacerbates class bias in education”.
Some students rely on AI for coursework because they are not taught the skills needed to succeed in that type of assessment.
Some students approach AI with feelings of fear and doubt: offput because it “feels like cheating”; lacking clarity on the rules and expectations of use; and the stress of time-pressured reasons to over-rely on AI.
AI and Assessment
The assessment categories are known to most students, but they have different ways of referring to them. They find them a helpful starting point for assessment clarity. Staff so sometimes offer contradictory guidance. In some cases, PGR teaching assistants encourage use when unit directors clearly forbade any AI use.
Students were sometimes aware of institutional guidance on AI, but many were not and there was a repeated call for more training and guidance on AI within degrees. There were also concerns about what defines AI and AI use: “Lots of tools have AI by default, so it’s hard to define what is AI”.
“If AI is used in an assessment, then AI skills should be set” – for this student, they see some assessment design as setting students up for failure when the assessment requires AI use, but AI literacy is not developed to match that requirement.
Students don’t see the point of in-person essay writing to avoid AI misuse, as it is “just regurgitating”.
Time pressure is a reason given for much AI misused or over-reliance in coursework. Students wonder if there are ways to learn time management with AI to mitigate this and think about how to spread work out over longer periods of time.
Further assertions and reflections on AI in Education
- Students recognise that AI is used differently in each subject, and this can make the education landscape uneven. What’s right in one subject isn’t perfect for another. The inconsistency across Schools does not always land positively, however, especially across cognate disciplines.
- Some students lament the lack of any engagement with the topic of AI in their discipline and are frustrated there is no discussion within their degree – that a Category 1 ban on all AI is the only response to the subject.
- Teachers’ opinions (pro or con) “really influence us”.
- Students don’t like the idea of using AI instead of staff members (to cut costs).
Get in touch to discuss AI further, ai-education@bristol.ac.uk; add your reflections on what this means for policy and practice below.
Leave a Reply