Design a practical skill assessment with AI
Set a realistic task, decide what good work looks like and ask learners to explain a decision. Use more than one observation before drawing conclusions.
Design a practical skill assessment with AI if the usual quiz rewards vocabulary more than performance. The skill-assessment guide starts with an activity that resembles the work you need done. Someone who understands sorting can demonstrate it on a small order table and explain what should happen when two orders share a date.
Decide what counts as a good result before seeing the answer. Change one condition afterwards to test whether the learner can carry the idea into a new situation. A fluent explanation is one piece of evidence. An independent task or another observer can reveal something the first attempt missed.
What it does
- Set a task that reflects the required level of performance.
- Write criteria before seeing the learner’s submission.
- Ask for the reasoning behind an important choice.
- Change a condition to see whether the method still holds.
- Give the learner feedback based on visible evidence.
Install and use
- Install the skill in your agent and describe the work, expected level and available time.
- Ask, “Design a short sorting task with one repeated date. Write the criteria and a follow-up question.”
- Compare the work and explanation against the criteria, then use another sample if the decision has consequences.
FAQ
Source
Made by Amey Thakur and Archit Konde. Original skill: skill-assessment, published on GitHub under the MIT licence.
View on GitHub