Reflective journal My account of the SDF work, written in February 2026 about events in 2025.
Technical materials The code, processing logic and diagrams of what I built, showing how it worked.
Dated messages Messages from the time, used only to confirm dates and the order of events.
Participant account What an interviewee described. Not observed or checked independently.
Interpretation dashed outline Conclusions drawn from the evidence.
Stage of use Whether a change was in use, being tested, explored or proposed. These labels describe a stage, not whether the change succeeded.
Setting 3 · Breadth · Chapter 6
Eight staff described using GenAI in learning design, teaching, research, research support, student coordination and strategy-facing work. GenAI helped as a starting point, for comparison, summaries and troubleshooting. Whether it worked depended on the task, the tool, the source material and the person's judgement.
Professional staff · learning design
Reviewed authoritative-sounding draft learning material with academic colleagues supplying subject knowledge. In one episode, unsupported claims took comments, emails and meetings to resolve.
“The co-pilot has such an authoritative voice”
Academic staff · teaching and qualitative research
Read research material first, then compared GenAI output with it. In one manuscript, the output missed a broader interpretation that SP02 then supplied. Had been told that research-participant material could be used only in Copilot.
Casual academic · teaching and assessment coordination
Coordinated short student meetings across Canvas, Outlook, Teams, calendars and local booking methods, and proposed a Copilot agent to help with bookings. Separately, used convincing but incorrect course-code information before correcting it after checking with students.
Change described: proposedAcademic staff · teacher education
Used GenAI as a starting point for syllabus alignment, themes in papers and mapping tables, then checked it against the syllabus and teaching purpose. Reported substantial time savings on some synthesis tasks.
“Because it gives me a springboard. I use it as a springboard. It's like a starting point”
Professional staff · research support
An early classification trial gave plausible explanations with invalid codes. Later reported a classification agent constrained to selected sources and a grant-statement review agent used by colleagues. Explored an external impact-case tool on a limited basis.
Change described: reported in use · exploredAcademic staff · teaching and subject coordination
Lost about 20 minutes to a desktop Teams problem that Copilot neither caused nor solved. Built an attendance workbook with Copilot's help on formulas, used across subjects. Caught one student's comment appearing in other students' feedback drafts.
Change described: reported in useProfessional staff · strategy-facing work
Received Copilot after the initial pilot and learnt mainly alone, with no set training time. Used it for searching, comparing documents and summaries; spreadsheet work stayed tentative and advanced reporting and agent ideas were not implemented.
“So I think what's missing for me is more of a community of practice”Change described: proposed
Academic staff · research and grants
Research-software errors could stop work; SP08 put error messages into Copilot to understand them, then still had to check the fix worked. Gained access to Copilot after not receiving it in an initial allocation.
“you can put the error message straight into Copilot and it tells you exactly what the error is and how to fix it”Change described: reported in use
Roles, access and experience are as participants described them, generalised to protect their identity. Licence tiers and permissions were not independently checked. These accounts do not show how common any practice is across the university. Thesis Sections 6.1–6.2
A recurring pattern across different settings, not one shared workflow; not every output went through every step. Adapted from Thesis Figure 6.6.1. Thesis Section 6.6
Timing of access
SP07 received Copilot after the pilot and trained themselves; SP08 gained access later, after not being included in an initial allocation.
Licences and permissions
SP05 said sharing agents was limited by colleagues' licences; SP02 had been told research material could only be used in Copilot.
Staying responsible
SP04 checked teaching material against the curriculum; SP06 decided students' results before using Copilot to organise feedback, then checked it.
Access, permissions and opportunities to learn differed between participants. These descriptions reflect their experiences. Thesis Section 6.8
RQ1 · Friction
Some friction came from other systems (Teams, scattered booking tools, research software) and some from generated output (wrong course codes, invalid codes, feedback carried between students).
RQ1 →RQ2 · Responses
Agents, a workbook and troubleshooting were reported in use, alongside exploration and proposals. The accounts give too little detail about how they developed to show workaround formation.
RQ2 →RQ3 · Continuing work
Checking, correcting and learning sat behind the outputs, with mixed consequences. The accounts do not support a general finding of invisible labour.
RQ3 →