Training Agenda (Sunday 10 May 2026)
Training Objective: Build concrete supervisory and methodological capability across the full research lifecycle, with AI as a central—but controlled—tool. This is not a lecture about AI. This is a working day where you will use AI, critique AI, and leave with both sharper skills and a shared institutional stance on what responsible AI-mediated supervision looks like.
Training Host: Dr. Jim Wagstaff, GSBM Faculty, Educator & AI Innovator, Singapore
09:00–09:15 | Calibration: Where Are We Starting From?
AI experience in the room will vary. Before we can build together, we need to know what we’re working with.
- Hands-on pulse check: Everyone completes the same task in ChatGPT, Claude, etc.
- What did you notice? What surprised you?
- Framing the day: You will experience AI as your candidates will, then judge it through the lens and experience that a supervisor must
Outcome: A shared baseline of experience. No assumptions. No one left behind or held back.
09:15–10:45 | Applied Lab I
AI Across the Research Lifecycle
Your doctoral candidates will use AI. The question is not whether but how—and whether they will use it well or badly. This lab puts you in the shoes of the doctoral candidates.
Station A: AI and the Literature (30 min)
- Tools: NotebookLM, Claude, ChatGPT, etc.
- Upload papers. Generate synthesis. Query across sources. Create an audio overview.
- The hard question: What does AI surface that you might have missed—and what does it flatten, oversimplify, gloss over, or get wrong?
Station B: AI and Qualitative Analysis (30 min)
- Tools: Claude, ChatGPT, NotebookLM
- Feed in interview transcripts. Ask for thematic coding. Request pattern identification.
- The hard question: How does AI’s coding compare to your instincts? What would you accept, revise, or reject?
Station C: AI and Quantitative Analysis (30 min)
- Tools: Claude, ChatGPT, Code Interpreter
- Work with real data outputs—regression tables, survey results
- The hard question: Could a candidate use this to understand their data better—or to avoid understanding it altogether?
Outcome: Direct experience with AI’s capabilities and limitations. Initial instincts about where AI helps and where it misleads.
10:45–11:00 | Break
11:00–12:00 | Applied Lab II
AI, Mixed Methods, and the Integration Problem
Mixed methods research is where AI gets interesting—and dangerous. Integration requires judgement that AI can simulate but not possess.
- Scenario: A candidate has qualitative themes AND quantitative survey data on the same phenomenon
- Task: Use AI to explore integration—joint displays, convergence and divergence, narrative synthesis
- The hard question: Where might AI manufacture false coherence? What must a supervisor probe?
Outcome: Understanding of AI’s potential and pitfalls in the most complex methodological territory your candidates will navigate.
12:00–12:30 | Supervisor Reflection
When to Say: “Read the F***ing Papers”
Now put on your supervisor hat. Review outputs from the morning—yours or a colleague’s—as if a candidate submitted them.
- What would you want to know about how this was produced?
- What capability must the candidate still demonstrate themselves?
- What’s your green light, yellow light, red light?
Human judgement is the non-automatable core of supervision. This session is about locating exactly where that judgement must be applied.
Outcome: The shift from experiencing AI to supervising its use. Surfaced tensions that will inform the afternoon.
12:30–13:30 | Lunch
13:30–14:15 | Data Clinic
Ethics, Integrity, and Real-World Data Problems
AI does not solve the fundamental problems of research integrity—it complicates them. This session addresses what supervisors must enforce regardless of the tools candidates use.
- Data management failures: What goes wrong, and why it matters
- Auditability and reproducibility: Can you trace how the results were produced?
- AI-assisted data collection: Code generation, scraping, survey automation—where are the risks?
- Institutional responsibility: What must the institution guarantee, and what falls to the supervisor?
Outcome: Clear understanding of the ethics and integrity issues that AI amplifies but does not create.
14:15–15:00 | Writing & Integrity Lab
AI-Supported Writing Without Academic Fraud
The difference between editing and authoring matters. So does the difference between using AI to clarify your thinking and using it to avoid thinking altogether.
- Editing vs. authoring: Where is the line?
- Detecting over-automation: What does AI-written prose look like, and why does it matter?
- Making authorship defensible: What documentation and transparency should candidates provide?
Hands-on exercise:
- Use AI to draft, summarise, or translate findings for non-academic audiences
- Compare: What happens when AI drafts versus when AI refines your draft?
- Pair critique: Swap outputs. Score on accuracy, insight, originality, voice.
Outcome: Practical capability in AI-supported research communication. A critical eye for detecting over-automation and defending authorship.
15:00–15:15 | Break
15:15–16:30 | Synthesis Workshop
What Supervision Must Now Enforce
AI will be used regardless of policy. The question is whether your institution has a defensible, transparent stance—or leaves every supervisor to figure it out alone.
Gallery walk (15 min): Review the tensions and insights surfaced throughout the day.
Sorting exercise (35 min): Working in groups, categorise AI use across the research lifecycle:
| Research Phase | Encouraged | Permitted with Transparency | Discouraged or Prohibited |
| Literature synthesis | | | |
| Qualitative coding | | | |
| Quantitative interpretation | | | |
| Mixed methods integration | | | |
| Data collection & management | | | |
| Writing & communication | | | |
Draft guidance (25 min): Together, we will produce a working draft of AI guidance for supervision: what candidates should be told at the start of their journey, and what supervisors must enforce throughout.
Outcome: A co-created institutional asset. Not imposed from above—built from shared experience.
16:30–17:30 | Presentation of Findings
From Working Groups to Shared Perspective
Each group presents their draft guidance to the full room.
- Where do we align?
- Where do we differ—and why?
- What tensions remain unresolved?
This is not about forcing consensus. It is about making differences visible so faculty can make deliberate choices rather than drift into inconsistency.
Outcome: A consolidated draft of AI-related guidance for doctoral supervision: co-created, debated, and owned by the people who will use it.
17:30–18:00 | Closing
Commitments: What Will You Do Differently?
- Which tools and use cases were most valuable? Least?
- One thing you will try with your next supervisee
- One thing you will watch out for
- Who owns the draft guidance? When do we revisit it?
Outcome: Personal accountability. Institutional ownership. A path forward.
Exit Condition
Participants leave Day 3 with:
- Experience using AI tools across the full doctoral research lifecycle—not watched someone else use them;
- Sharpened judgement about when AI helps candidates learn and when it lets them avoid learning;
- Clear understanding of ethics, integrity, and auditability requirements in AI-mediated research;
- A co-created guidance document that makes guidance explicit and useful;
- Personal commitments to supervise differently in an AI-mediated world.
What to Bring
- A laptop with internet access
- A Google account (for NotebookLM and Gemini)
- Access to ChatGPT, Gemini, Perplexity, Claude, etc. (free tiers work; paid tiers unlock more capabilities)
- Willingness to experiment, make mistakes, and think critically
