PREPARE FOR WHAT’S NEXT · WEEK ENDING 16 October 2026
Use AI without outsourcing your judgement
Build one capability this week: check an AI-assisted research interpretation against the evidence, then explain your own decision.
What is changing?
AI tools can help people complete a task, but completion and learning are different outcomes. A school mathematics experiment found risks when students used an unguarded assistant; a separate university physics study reported benefits from a deliberately designed tutor. These are different interventions and settings, not a simple verdict for or against AI.
Our practical interpretation
For students and practitioners, a useful preparation habit is to attempt a task independently, scrutinise assistance, and explain a decision without relying on the tool. This is our proposed UX activity, not an intervention tested by the cited studies. It develops a capability you can practise now without predicting which job titles will emerge.
Try this now
Use the fictional notes below. You can work on paper or in a document. An AI account is optional: a deliberately flawed example response is supplied.
- 01 · 10 minutes
Make your own reading
Read the fictional notes. Write two tentative themes and attach the note IDs that support each. Record one exception or uncertainty. Keep observations separate from your interpretation; these short notes cannot establish prevalence or causation.
- 02 · 10 minutes
Compare and challenge
Use the optional prompt with an AI tool, or audit the supplied example. For each claim, mark Supported, Uncertain or Unsupported and cite the relevant note IDs. Look for omitted detail, invented explanations and unjustified generalisations. You do not need to find an error if a claim is well supported.
- 03 · 10 minutes
Decide and explain
Close the AI response. Write a 100-word recommendation in your own words: what would you investigate next, why, and what remains unknown? Reopen your comparison to check the evidence. Save one decision you changed and one you retained, with reasons.
FICTIONAL PRACTICE MATERIAL
A small research exercise
Imagine reviewing early notes about a community workshop booking service. The six statements below are invented snippets from separate fictional people. They are a small teaching example, not a transcript, representative sample or complete thematic analysis.
- N1
I booked on my phone during a break. I had to scroll back up to check the workshop date before paying.
- N2
The booking form was fine on my laptop. I stopped when I saw the cancellation policy; I could not tell whether I would get a refund.
- N3
With my screen reader, the calendar buttons were announced as ‘button’. A friend helped me choose a date.
- N4
I wanted to reserve two places, but I could not tell whether the price shown was for one or both.
- N5
I booked successfully on my phone. The confirmation email arrived quickly and clearly showed the date.
- N6
I looked for a workshop after work. All the available sessions were in the afternoon, so I left without starting a booking.
Optional prompt for an AI comparison
These are invented notes for a learning exercise, not real participant data. Suggest two tentative themes. For every claim, cite the note IDs supporting it and identify any counterexample. Separate observation, interpretation and proposed action. Do not invent quotations, demographics, frequency estimates or causal effects. State what cannot be concluded from these notes. Finish with one question for further research.
No AI account? Audit this example response
All users struggled with mobile booking because the checkout is too long. The main problem is slow confirmation emails. Shortening the checkout and sending faster emails will increase bookings. Accessibility was not a concern, and everybody wanted evening sessions.
After your attempt: checks to consider
- N1 describes a mobile difficulty, while N5 describes a successful mobile booking. ‘All users struggled’ ignores that counterexample; checkout length was not measured.
- N5 reports a quick confirmation email. No note supports slow emails as the main problem.
- N3 raises a specific accessibility issue. N2 concerns refund clarity; N4 concerns price clarity. Do not collapse different barriers into a single explanation.
- N6 expresses a scheduling need. It does not show that everybody wants evening sessions. The notes cannot establish that a proposed change will increase bookings.
- Possible themes include clarity at decision points and access to suitable bookings, but other interpretations may be defensible. A theme needs traceable support and preserved differences, not agreement with a model answer.
SOMETHING TO SHOW
Keep a record of your judgement
Keep a one-page evidence record: your initial themes, the comparison with source IDs, your final recommendation and its limitations. Label it ‘fictional learning exercise’, record the tool and date if used, and disclose the assistance. In a portfolio or interview, explain your decisions rather than presenting this as client or participant research.
For experienced practitioners: Ask a colleague to challenge your evidence links and proposed next step. Compare disagreements before considering a team review checklist. Do not turn this small exercise into a benchmark for AI reliability or a claim about research quality in general.
Supporting readings
PEER-REVIEWED JOURNAL · BACKGROUND
Generative AI without guardrails can harm learning: Evidence from high school mathematics (opens an external website)
AI support improved practice performance, but unguarded assistance could undermine later performance without AI. A tutor using teacher-informed safeguards mitigated the harm.
For this activity: Assess what you can explain independently after using assistance, as well as what you can produce with it.
Evidence, access and limitations
Peer-reviewed randomised field experiment comparing GPT-4-based support with and without learning safeguards and a no-AI condition in high-school mathematics.
Reviewed: Publisher abstract, significance statement and study-design excerpt; full methods not appraised.
Limitations: School mathematics is not UX research. Transfer to qualitative interpretation or professional practice has not been established here. This brief is based on selected publisher sections, not a full methodological review.
Opens the publisher outside this reading room. If nothing opens, copy the source address below.
Reference tools · Paperpile & more
Using Paperpile? Open the original source above, then use the Paperpile extension to save it to your library. Its inline buttons appear on supported websites.
Paperpile’s browser extension guide (opens an external website)Or download the original reference and import the file into Paperpile, Zotero or EndNote. Check the details against the source before citing.
Citation details are requested from Crossref when you download. The file contains a reference, not the full paper.
PEER-REVIEWED JOURNAL · BACKGROUND
AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting (opens an external website)
The study reported stronger learning outcomes from a deliberately designed AI tutor than from the comparison active-learning lessons. The tutor incorporated pedagogical guidance rather than offering unrestricted chat alone.
For this activity: Consider how support asks you to reason, rather than judging its usefulness only by speed or polished answers.
Evidence, access and limitations
Peer-reviewed randomised controlled comparison of a purpose-built AI tutor and an active-learning class in undergraduate physics.
Reviewed: Publisher abstract and indexed results/discussion excerpts; full methods and supplements not appraised.
Limitations: A specific tutor, course and short intervention; this does not establish that generic chatbots outperform teaching or support complex UX judgement. The authors explicitly caution against assuming superiority in every context.
Opens the publisher outside this reading room. If nothing opens, copy the source address below.
Reference tools · Paperpile & more
Using Paperpile? Open the original source above, then use the Paperpile extension to save it to your library. Its inline buttons appear on supported websites.
Paperpile’s browser extension guide (opens an external website)Or download the original reference and import the file into Paperpile, Zotero or EndNote. Check the details against the source before citing.
Citation details are requested from Crossref when you download. The file contains a reference, not the full paper.
POLICY REPORT · BACKGROUND
Skills in the AI age (opens an external website)
The report examines changing skill needs and policy responses to AI. It emphasises that employment effects vary across sectors, places and skill levels.
For this activity: Use the report to frame questions about your intended sector, then identify a capability you can practise and demonstrate now.
Evidence, access and limitations
Institutional policy synthesis about AI, changing skill requirements and labour-market responses; not an experimental study.
Reviewed: Official publication abstract and metadata; full report not appraised for this feature.
Limitations: Broad labour-market context, not a UX hiring forecast or proof that one course, tool or certificate secures employment. This brief reflects the abstract, not an assessment of all underlying evidence.
Opens the publisher outside this reading room. If nothing opens, copy the source address below.
Reference tools · Paperpile & more
Using Paperpile? Open the original source above, then use the Paperpile extension to save it to your library. Its inline buttons appear on supported websites.
Paperpile’s browser extension guide (opens an external website)Or download the original reference and import the file into Paperpile, Zotero or EndNote. Check the details against the source before citing.
Citation details are requested from Crossref when you download. The file contains a reference, not the full paper.
UX Observatory · Liz Coulter-Smith · Prepared with OpenAI assistance. This editorial activity is not a validated training intervention, qualification or guarantee of employment.