Skip to content
All quizzes

Self-Assessment · ~6 min

The AI/UX Maturity Model

Assess your team's capability to deliver exceptional AI experiences.

20 clear questions to benchmark your organization's AI UX maturity. Rate each one, and the result will recommend the exact three practices your team should prioritize next.

0 of 20 answered

0%

Level 1 · Level 1: Emerging Practice

AI is introduced as an additive feature. Quality control relies on post-launch monitoring.

  1. AI failures are caught in production by users, not internal QA.
  2. We retrain or fine-tune the model only after a complaint reaches us.
  3. We do not have a written list of what the AI is supposed to refuse.
  4. There is no on-call rotation for AI-specific failures.

Level 2 · Level 2: Developing Practice

Quality and failure tracking are formalized. Pre-release testing is standard practice.

  1. We do post-mortems when an AI failure causes a customer-visible incident.
  2. We track AI-specific bug reports separately from non-AI bugs.
  3. We have a documented escalation path for AI-related complaints.
  4. We retrain the model on a fixed cadence, not just on incidents.

Level 3 · Level 3: Established Capability

Safety and error recovery are proactive. The team anticipates edge cases.

  1. Every model change runs through an evaluation suite before it ships.
  2. Designers review prompts as part of the release checklist.
  3. We have written acceptance criteria for AI behavior, owned by a product person.
  4. We can roll back a model change in under an hour.

Level 4 · Level 4: Advanced Integration

UX decisions are driven by continuous user feedback and trust metrics.

  1. We measure user trust signals (corrections, undos, abandonments) and act on them weekly.
  2. We run shadow models against production traffic before any major model swap.
  3. The product surfaces its own confidence and the user can see the model’s uncertainty.
  4. We have a Behavioral Contract artifact that engineering and design both reference.

Level 5 · Level 5: Industry Leader

The product anticipates intent and aligns perfectly with user goals.

  1. The product slows down on classes of action where users frequently reverse the AI.
  2. We allocate part of the recommendation surface to novelty so the user is not boxed in.
  3. We run an honesty cohort that receives a less-personalized experience as a control.
  4. The team can name the dark patterns it refuses to ship and has refused them in the last quarter.