Last updated: 2026-11-06
Lab 5: AI-Proof Skills Audit
This closing lab turns the series back on its own participants. Having spent four sessions on frames, feedback, wicked problems, and emergence, the question now is personal rather than abstract: which of your own current skills sit on the tame side of the line, where a model is already strong and getting stronger, and which sit on the wicked side, where framing, judgement, and governance still have to be done by a person? Treat this as a risk audit, not a confidence exercise — the same likelihood/impact discipline the series has applied to a frame collision, a runaway loop, and a wicked problem, aimed this time at your own career.
Learning Objectives FoundationalKnowledge that endures for decades — core principles
- Audit your own current skill set against the AI-strength / human-value distinction from the series.
- Distinguish "AI did the typing" from "AI did the thinking" in a piece of your own recent work.
- Produce a concrete, falsifiable plan for which skills to deliberately strengthen before graduating.
Pre-Lab Reading FoundationalKnowledge that endures for decades — core principles
Read AI-Assisted Development, Learning as a Feedback Loop (and the AI Partner), and the "AI and the Future Software Engineer" section of Software Meets Reality. Bring one piece of your own recent work where you used an AI assistant — a coursework submission, a personal project, a lab exercise from earlier in this series.
Materials
- The capability table from Software Meets Reality (Coding, Debugging, Programming, Architecture, Systems Thinking, Governance, Framing Problems), printed or displayed.
- The self-check table from Learning as a Feedback Loop ("Could you re-derive the solution without the transcript?" and the rest).
Session Plan (120 minutes) Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice
| Time | Block | Activity |
|---|---|---|
| 0:00–0:15 | Briefing | Recap the capability table and why "AI strength" and "human value" in it are an argument, not a measurement. |
| 0:15–0:45 | Personal case audit | Working alone: run your own brought-in example through the learning-loop self-check table, honestly. For each row, write which answer you'd actually give, not the one you'd like to be true. |
| 0:45–0:55 | Break | — |
| 0:55–1:20 | Capability self-rating | In pairs, rate your own current ability against each of the seven capabilities in the table (not against the AI — against where you personally stand right now, low/medium/high). Your partner's job is to challenge any rating that isn't backed by a specific example from the last twelve months. |
| 1:20–1:35 | Score the exposure | For each row where your own rating is low and the row sits on the AI-strength side of the table, score the career risk using the same likelihood/impact logic as Risk Management & Mitigation: how likely is it that this specific gap matters within the next two years, and how bad is the consequence if it does? |
| 1:35–2:00 | Strengthening plan | Identify the one capability on the human-value side of the table (architecture, systems thinking, governance, framing problems) where your own rating is lowest. Write one concrete, checkable action you'll take before the end of this course to move it — not "get better at systems thinking," but a specific exercise, module choice, or project decision with a date attached. This is your mitigation, in the same Avoid/Reduce/Transfer/Accept sense as every other lab in this series — name explicitly which of the four it is. |
Deliverable Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice
Your completed self-check table, your seven-row capability self-rating with one supporting example per row, the exposure scores for your lowest-rated AI-strength rows, and the single dated strengthening action — with its Avoid/Reduce/Transfer/Accept label — from the final block.
Related Topics
- Software Meets Reality: Systems Thinking, Frame Analysis, and the Future Software Engineer — the argument the whole series has been building toward.
- AI-Assisted Development — the verification checklist this lab's self-check builds on.
- Learning as a Feedback Loop (and the AI Partner) — the self-check table used in the first block.
- When the AI Becomes the IDE — the fuller argument for why specifying a problem precisely is the skill that survives.
- Risk Management & Mitigation — the likelihood/impact scoring and Avoid/Reduce/Transfer/Accept framework this lab applies to your own skill set rather than to a project.
- Lab 4: Emergence Hunting — previous in the series.
References
-
Barbini, U. (2026). Process Over Magic: Beyond Vibe Coding — Faster, Smarter, and Safer Coding with AI Assistants. The Pragmatic Bookshelf.
Graduate Attributes
Directly developed
- Self-Directed Learner: Readers plan a single dated strengthening action based on their self-rating.
- Reflective: Readers examine their own assumptions by rating their capabilities against specific examples.
Modelled on this page
- Subject Specialist: Disciplinary concepts such as tame and wicked problems are applied to real-world software engineering practice.
- Critical Thinker: Judgement is used to distinguish between skills where models are strong and those requiring human governance.
University of Reading Graduate Attributes framework.