As designers, we want to know how we can leverage these new AI/ML capabilities to solve problems, fulfil user needs, and create meaningful experiences. Although there's no simple 3-step process, this page provides some ideas to start spotting opportunities.

Starting points

Experience design opportunities for AI might arise from a few different starting points:

  1. Starting from user need (!) Exploring if and how ML capabilities might support in solving a user / business challenge - e.g. decision making, customer service reply times.

For inspiration: just scroll down to the next big header.

![<https://s3-us-west-2.amazonaws.com/secure.notion-static.com/e662fb76-83fc-498f-ad8e-2e9daa200ec4/Screen_Shot_2019-05-20_at_16.16.04.png>](<https://s3-us-west-2.amazonaws.com/secure.notion-static.com/e662fb76-83fc-498f-ad8e-2e9daa200ec4/Screen_Shot_2019-05-20_at_16.16.04.png>)

  1. Starting from data available Wondering what insights or value might be derived from existing, company-owned data - e.g. user usage data, sets of images or user-generated text.

  2. Starting from existing applications

    Interpreting the value of existing applications seen in other products / services for one's own context - e.g. recommender systems for content, anomaly detection for fraud.

    For inspiration, check:

Interesting examples + case studies of AI across industries

  1. Starting from new capability

Finding valuable applications and use cases for available algorithms or ML capabilities - e.g. voice sentiment analysis, psychographic personas.

For a bunch of brainstorm prompts translating AI/ML capabilities in more abstract, user-facing and designerly terms, check:

What-if machine learning brainstorm prompts

Spotting AI/ML opportunities

Map the customer journey as you usually do, then look for parts of the process that:

  1. could be automated or augmented. What are you asking your users to do that you could do for them? How can your product / service support assistive / agentive / automated modes?
  2. are delayed or sequenced. Where are you making your user wait? Where might your user benefit from real-time feedback to their queries?
  3. are isolated. Where might your user benefit from context-relevant understanding? How can your product / service take advantage of the device seeing and hearing the physical world your user moves in (almost experiencing the world through their eyes)?
  4. are strenuous / ill-fit to screens. Where might your user benefit from another input method than tapping and typing on their screen?
  5. are generic. Where might it be valuable for the system to adapt to the user over time? What insights might you be able to draw from the data points you have?
  6. are habitual. Which patterns or combinations of behaviour are so frequent you might be able to predict what the user will want to do next?
  7. are sensitive. Where might your user benefit from the system being emotionally-aware and affective?
  8. are labor-intensive. Where might you automate part of the process? How could you improve the experience with 10 'dumb' (artificial) interns?