Even for AI, there are considerations more important than technical feasibility

We started the year in a frenzy of AI/ML interest and it’s still going strong. It seems like every event we attend has an AI/ML panel, and our #machine-learning Slack channel regularly discusses a new article, open-source library or resource. In reflecting on what it takes to build a good AI solution, our best practices remain the same and starting with a quality foundation is even more important.

AI has tremendous potential to make improvements for humanity, especially in education, learning and health verticals. For a lot of folks, however, they have a limited understanding of how it works, and the result is these solutions can be met with apprehension and even fear.

It (still) starts with quality Design & Research

Designing good products takes research, and not just a few googles. You need to conduct quality research that lets you deeply understand who your target users are, their goals, needs, and challenges. When it comes to AI powered solutions, understanding a user’s perception of their interactions and what’s happening behind the scenes is even more important and a key detail in meeting your adoption goals.

To be able to do this, you must approach research with empathy, and have the ability to understand and investigate the emotions of the user at each step. Empathy is the ability to understand and share the feelings of another person. When building software, like a website or mobile app, the team at thoughtbot strives to use empathy-driven design at each stage, including the earliest brainstorming session through the launch planning.

User feelings and perceptions are more challenging to capture and extrapolate. Even if the product is fairly simple, and usability testing informs you of success rates for workflow completion, being able to successfully understand how users feel throughout is paramount, especially in AI based products. Do they trust the product’s recommendation? Are they fearful of their data being captured in a way they don’t understand?

How do I get started?

If you haven’t conducted user research before, we have a number of resources to check out. When it comes to capturing user sentiment for AI/ML software, here are a few tips from our Design team and some industry experts on ways to get started:

  1. Determine who your customers are and be specific. Use these exercises to generate a weighted list of market segments you can conduct research with.
  2. Now that you know who you should be talking to, spend some time with your customers. This could be in many forms like shadowing sales meetings, customer interviews, talking to churned customers, conducting usability testing, or sitting alongside the support team for the afternoon. Shift focus to doing one thing: empathizing with the customer.
  3. If you go the customer interview route, set yourself up for success before beginning your interviews, which includes writing a script.

The risk you face by “skipping” the research step, is later not knowing what’s holding you back or lacking the insight to identify the right next feature or large pivot to take.

Here to Help

Doing user-first research, and taking a design-led approach to build the right solution the first time is by far our go-to starting point for building new products or outlining the right improvements. If you need help with Usability Testing or Customer Discovery work, we are on stand-by to help and can supercharge your strategy in as little as 2 weeks. If you are interested in AI/ML but aren’t sure what makes sense for your business, we have Discovery sprints tailored to sussing out use cases and testing their feasibility.