Evaluating human–AI interaction
Look beyond the quality of an AI answer. Explore how people understand, check and act on it.
What you’ll be able to do
- Map the user's role and the AI system's role
- Identify situations where misplaced trust could matter
- Design tasks that test checking, correction and recovery
- Separate model performance from interaction findings
Who this is for
UX practitioners, product managers and teams introducing AI-assisted features.
Before you join
A working familiarity with digital products and basic UX concepts will help you participate. Share your experience when enquiring so we can recommend the right starting point.
Inside the programme
Frame the human–AI task and boundaries
Frame the problem, discuss examples and make the key decisions explicit.
Explore uncertainty, trust and reliance
Apply the method to a guided example and compare approaches with peers.
Create scenarios with plausible AI mistakes
Practise the skills, review what happened and refine your approach.
Plan an evaluation of oversight and recovery
Bring the learning together and plan how to use it in your own context.
Learning by doing
Evaluate an example AI assistant using deliberately imperfect outputs in a simulated task.
What you take away
A human–AI task map, scenario set and evaluation planning canvas.
A realistic application
An AI assistant drafts a service summary. Design a study to understand whether users verify source information, notice an incorrect statement and recover from the mistake.
What we review together
Review the division of responsibility, scenario realism, checking behaviour and the limits of the proposed evaluation.
After the programme
Build an agreed set of oversight scenarios and test them in a controlled setting with appropriate users.
Bring the learning into a live engagement
For teams that want guided application, a separate audit or research engagement can connect the training to an actual product question. We agree the scope, access and deliverables before work begins.
Explore research and audit services
Feedback and completion
Learning is reviewed through practical exercises and discussion. Any completion record confirms participation in The UX Faculty programme; it is not a professional licence or an externally accredited UX certification.