Summary
Guided practice asks learners to perform a bounded task themselves while a facilitator, artifact, or workflow structure provides scaffolding.
For this project, guided practice is where participants begin to use AI with judgment instead of merely watching someone else do it.
Evidence status
Assessment: evidence-backed
Primary support:
- Source - The ICAP Framework (Chi & Wylie, 2014)
- Source - Teaching and Assessing for Transfer (National Research Council, 2012)
Why this pattern belongs here
- Passive exposure is not enough for transfer into live work.
- AI-assisted workflows are easy to imitate superficially unless learners must make their own decisions.
- Adult professionals need practice that resembles real job conditions if the training is meant to change real behavior.
What this pattern is trying to achieve
- convert examples into action
- make decision-making visible under mild pressure
- surface misunderstandings early
- let facilitators correct unsafe habits before they harden
When to use it
- after a worked example
- when introducing a workflow that needs judgment rather than rote memory
- when the desired outcome is job transfer rather than awareness
When not to rely on it alone
- when learners lack enough framing to start safely
- when the task is too large or ambiguous for the current session
- when there is no feedback loop after the exercise
Patterns and practices
- make the task realistic but bounded
- give a clear objective and known constraints
- provide prompts that require explanation and verification, not just output
- stop learners at key checkpoints and ask what they trust and why
- keep facilitator feedback close to the moment of practice
- use pair or small-group discussion when useful
Good forms for this project
- guided debugging exercise
- test-case expansion exercise
- architecture-option critique exercise
- ambiguity-detection exercise for acceptance criteria
Anti-patterns
- giving a task that is too open-ended for the time available
- grading only the final answer instead of the reasoning and review path
- providing so much scaffolding that no real judgment is required
- leaving learners without feedback until long after the exercise
Example application in AI enablement
Give engineers a bounded defect and ask them to use AI to generate hypotheses, then require them to identify what must be verified manually before declaring a fix.
What should accompany this pattern
- prior framing or demonstration
- reflection and debrief
- follow-up retrieval or real-work application soon after