What to check when hiring for ai interface design
For AI product UX, assess how the interface communicates generated output, uncertainty, sources, user control, and recovery when a system is wrong or unavailable. Distinguish a deployed product experience from a concept, prototype, or agency capability claim.
Questions to put in the brief
- How does the interface show generated content, uncertainty, and source context?
- Can users inspect, edit, reject, or recover from an AI result?
- What happens when the system is wrong, slow, or unavailable?
- Is the example a concept, prototype, limited release, or deployed product?
Evidence to compare
Request a walkthrough of the user journey around the AI output, not just a screen featuring a prompt. Relevant evidence shows how users provide context, assess results, correct errors, retain control, and understand the system’s limits.
A common evidence gap
An AI-branded portfolio piece may be a concept rather than a product in use. Confirm deployment status and distinguish interface design from model development, safety evaluation, and measured user outcomes.
For a broader comparison framework, read What good agency evidence looks like.