# Using AI for Workplace Learning: A Trainer-Led Practice Example

Try a bounded AI practice exercise in workplace training and keep the trainer responsible for checking examples, feedback, and learner progress.

- Canonical URL: https://www.harmate.com/en/blog/ai-and-education-knowledge-is-available-direction-is-missing
- Author: Harmate Team
- Published: 2026-05-19
- Updated: 2026-10-19T22:58:55.717039+00:00
- Language: en

## Content

# Using AI for Workplace Learning: A Trainer-Led Practice Example

An AI tool can help create another practice scenario, but it cannot tell a trainer whether the exercise fits the learners, reflects current procedures, or demonstrates competence. A useful design gives learners a task to practise and keeps a qualified person responsible for checking the materials and feedback.

The U.S. Department of Education’s [report on AI and the future of teaching and learning](https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf) recommends keeping humans involved in educational decisions and describes educators’ role in reviewing AI-generated materials. UNESCO’s [guidance on generative AI in education and research](https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research) likewise frames the use of AI around human agency and pedagogical design. The example below applies those broad principles to one workplace-training exercise; it is not evidence that AI improves learning outcomes.

## Example: practising a customer-service handoff

A trainer is preparing learners to hand off a customer issue to another team. Instead of asking an AI assistant to “teach customer service,” the trainer gives it a narrow, fictional task: draft three short scenarios in which a customer’s request needs escalation. Each scenario should include the issue, what the first agent has already tried, and one piece of information the receiving team needs.

The trainer reviews every scenario before the exercise. They check that the scenarios match the organization’s actual handoff procedure, avoid invented policy details, and vary the context without introducing stereotypes. If the assistant generates an incorrect or confusing example, the trainer edits or discards it.

Learners then practise writing a handoff note for one scenario. The AI can offer a suggested note structure or ask a follow-up question, but it does not decide whether a learner passed. The trainer compares the work with a short rubric prepared in advance: Is the issue stated clearly? Is the action already taken recorded? Is the next team given the information it needs? The trainer discusses missing or unclear details with the learner.

## Keep the learning goal visible

A simple exercise can follow five steps:

1. **Set the task:** identify the work action learners should practise.
2. **Prepare the material:** use fictional details where possible and review every AI-generated scenario.
3. **Let the learner attempt it:** ask for a response before showing a model answer.
4. **Give specific feedback:** connect comments to the task criteria and let the learner revise.
5. **Ask for reflection:** “What did you change after the feedback, and what would you check in a real handoff?”

This creates a practice loop the trainer can observe. It does not establish that a learner can perform the task in every live situation; the trainer may need a separate demonstration in the relevant work context.

## What the trainer must check

**Accuracy:** Is the scenario consistent with current procedures and terminology? Remove invented facts instead of asking learners to treat them as plausible.

**Usefulness of feedback:** Does the comment point to an observable feature of the work, or does it make a vague judgement about the person? Prefer “The next team does not know which step you already tried” over “You are not detail-oriented.”

**Fair access:** Can learners who need another format, more time, or a non-AI alternative complete the practice? Use the course’s normal accessibility and accommodation process.

**Information handling:** Do not paste real customer records, employee details, or confidential cases into a tool unless the organization has explicitly approved that use. A fictional scenario is usually enough to practise the skill.

**Learner understanding:** Tell participants when AI contributed to an exercise and what the trainer will assess. Let them ask questions and correct a scenario that does not reflect the task.

## Review the exercise after use

After the practice, record which parts helped learners attempt the intended action, which instructions caused confusion, and what the trainer changed. Keep learner work and feedback within the training process; do not convert one exercise into a general score of a person’s ability or potential.

The [active-learning guide](/en/blog/active-learning-choose-an-activity-for-the-work-at-hand) offers a way to choose an activity from the action learners need to practise. The planning checklist in [Planning a Practical AI Pilot](/en/blog/shape-method-the-5-levers-to-succeed-with-ai-adoption-in-business) can help a team define a narrow test and a human review process before using an AI tool in a workflow.