Learn UX research methods for AI products that break our assumptions about how products behave.
A five-week live cohort that guides researchers through strategies for study design and execution on AI-powered products.
- Investment
- $495
- Format
- Live online · 5 weeks
- Cohort size
- Capped at 15
Lifetime access to recordings & materials
Revisit every session and resource, for good.
Full refund up to the first day of class
Change your mind any time before week one.
Certificate of completion
A shareable credential when you finish.
AI products are dynamic. UX research methods need to adapt.
Picture a usability test of a health coaching chatbot: your script assumes the bot returns a structured nutrition plan, exactly like it did in your pilot. Instead it outputs a list of stress-management techniques, and the session stalls. Your tasks assumed a static path, but the product has its own agency. These failures happen when researchers apply standard methods to dynamic systems. This course gives you a working set of methods and frameworks to design studies that account for unpredictable behavior, recruit and moderate participants who hold wildly divergent AI mental models, and evaluate trust and explainability as UX constructs, so your findings give product teams clear direction, not more ambiguity.
- 01
Diagnose what makes an AI feature require a different research approach
- 02
Design a study plan for an AI-powered feature
- 03
Recruit and screen for AI mental model diversity
- 04
Evaluate trust and provenance as UX constructs
- 05
Moderate research sessions involving AI features
- 06
Analyze and synthesize findings from AI product research
- 07
Frame and present AI research findings to product teams
→ Built for you if
- You're a UX researcher being asked to study AI features without having had formal preparation for the specific challenges this creates.
- You're a senior researcher at an organization doing AI product development, and you want to build a team-level research practice for AI contexts.
- You already run interviews, usability tests, or surveys as part of your normal practice and want to adapt that practice for AI-powered products.
× Not the right fit if
- You're new to UX research. This course builds on standard methods (interviews, usability testing, surveys) rather than teaching them. Start with foundations first, then come back.
- You're looking for technical AI/ML training: model evaluation, prompt engineering, or how the systems work under the hood. This is a research-methods course; no AI technical background required.
- You're looking to learn to do UX research with AI tools. Check out our course catalog for our other AI + UX research courses.
Assumed going in
- Familiarity with standard UX research methods: interviews, usability testing, surveys.
- No AI technical background required.
- Have an AI-powered product feature in mind that you're researching or want to research. You'll use it as your working example throughout the course.
01
What makes AI product research different
Diagnose the specific research challenges an AI-powered feature creates, instead of applying a standard playbook that assumes a stable, deterministic product.
- Identify the five core challenges: non-determinism, mental model instability, trust as behavior, explainability demands, and the moving-target problem.
- Use the constructs of adaptation load, adaptation anxiety, and provenance to talk about user trust.
- Analyze a case study of a standard research approach applied to an AI feature and pinpoint exactly what breaks down.
Workshop: Analyze a case study of standard research applied to an AI feature and identify what breaks down.
Homework: Select an AI-powered product feature you're researching or want to research, and write a one-paragraph diagnosis of the specific research challenges it presents.
02
Designing studies for non-deterministic systems
Draft a full study brief for an AI feature: method selection, stimulus design, and a screener built for mental-model diversity.
- Choose between concept testing, usability testing, diary studies, and contextual inquiry for AI features.
- Design stimuli and prototypes that isolate meaningful variables even when AI output is variable.
- Write screener questions and a sampling strategy that intentionally recruits for AI mental-model diversity.
Workshop: Critique and redesign a study plan that has AI-specific weaknesses.
Homework: Draft a study brief for your chosen feature, including method selection rationale, stimulus approach, and a screener question set.
03
Trust, provenance, and the moderation challenge
Build trust- and provenance-probing sequences into a discussion guide, and moderate sessions where AI output varies between participants.
- Operationalize trust as a behavior, not just an attitude, and build that distinction into a study design.
- Measure provenance: what users need to know about how AI output was generated.
- Moderate sessions with variable AI output while avoiding the social-desirability trap.
Workshop: Run and debrief short mock sessions with AI feature scenarios (live moderation practice).
Homework: Develop a discussion guide for your chosen feature that includes at least one trust-probing sequence and one provenance-exploration sequence.
04
Analysis, synthesis, and making findings land
Synthesize findings across participants with divergent mental models and frame them for product teams without falling into the "users don't trust AI" trap.
- Find cross-participant patterns through mental-model segmentation and behavioral indicators.
- Handle contradictory findings when participant behavior and stated preferences point in different directions.
- Frame AI research findings as actionable product direction for teams who may over- or under-weight AI capabilities.
Workshop: Synthesize a provided dataset from an AI usability study and present findings to a simulated stakeholder panel.
Homework: Begin your capstone: complete your full study plan, including research questions, method, stimuli outline, screener, discussion guide, and analysis approach.
05
Capstone: a complete research plan for an AI-powered feature
Present a complete, portfolio-ready research plan for an AI-powered feature of your choosing, with written feedback from the instructor and peer critique from at least two cohort members.
- Frame your research questions and lay out your method and stimulus rationale for the feature you've been working with all course.
- Write a screener and a discussion guide that includes a trust-probing sequence and a provenance-exploration sequence.
- Build out a full analysis plan, the final piece of the research plan package.
Deliverable: A complete research plan document suitable for inclusion in a professional portfolio or for immediate adaptation to a real work project.
Victor Yocco, PhD
Victor Yocco is the author of Designing Agentic AI Experiences (CRC Press, August 2026), a practitioner's guide to the UX research and design challenges specific to agentic AI systems. He has spent more than 15 years in UX research, with the past three focused on the psychology of human-AI interaction. His current research centers on the use and adoption of AI in enterprise software used by millions of people daily. He also consults with large organizations on AI adoption, helping them address the behavioral barriers that slow AI uptake among high-performing teams. He is the author of Design for the Mind (Manning, 2016) and publishes regularly with Smashing Magazine, UXmatters, A List Apart, and UXPA's UX Magazine.
Five weeks total: four weeks of live instruction plus a capstone week, one 2-hour live session per week, plus weekly office hours.
The first 20–30 minutes introduce or extend a concept; the rest is applied: case study analysis, live moderation practice, stakeholder simulations, and peer critique on study plans and discussion guides throughout the course, not just at the capstone.
Every session is recorded and posted the same day. You won't fall behind, though the live discussion, moderation practice, and stakeholder simulations are where most of the value lives.
Is this only for researchers, or useful for PM/design too? +
The methods are built around a UX researcher's workflow, but designers and PMs who run their own research on AI features will get just as much from it. The core challenge (AI products behave probabilistically and change underneath you) applies to anyone studying how people use them.
What do I actually walk away with? +
A complete, portfolio-ready research plan for an AI-powered feature of your choosing: research questions, method and stimulus rationale, a screener, a discussion guide with trust and provenance probes, and an analysis plan. You'll get written feedback from Victor and peer critique from cohort members, and the plan is built to be adapted directly to a real project at work.
What's the real time commitment? +
One 2-hour live session per week plus weekly office hours, across five weeks (four weeks of instruction plus a capstone). Each week also includes homework, building toward that final research plan.
Do I need an AI/technical background to take this? +
No. The course assumes you can already run standard UX research methods (interviews, usability testing, surveys), and no AI technical background is required. The focus is on adapting research practice you already have, not learning how models work.
Enroll with confidence. Full refund right up to the first day of class, no form, no friction. Once it starts, lifetime access to every recording is yours to keep.
Bring rigor to research on a target that won't stop moving.
Enroll Now · Jul 23 cohort15 seats · enrollment closes when full
