If you aren’t already working on projects with artificial intelligence components, you will be soon! Examples of AI, robotics, and machine learning are everywhere.
As these emerging technologies become more common, it’s important for BAs, BA leaders, and product owners to consider how the requirements process differs for AI projects. Here are 4 things that are especially important for AI-related requirements:
- Customer Empathy: Deep understanding of the wants, needs, and behaviors of the customer is critical for high-quality AI solutions. BAs should understand the customer’s journey. And it’s not a one-time analysis, good BA work includes ongoing evaluation of the customer experience.
- Experiments: The requirements process for AI projects should include many iterations of hypothesizing and experimenting to test theories and get frequent customer feedback.
- Chunking: An effective requirements approach uses empathy and experiments to gather data insights to help teams identify and prioritize increments of value. When teams break the solution into small chunks that allow for short, continuous delivery cycles, they get frequent feedback from customers. These feedback loops prevent teams from spending lots of time and money on the wrong stuff.
- Avoiding Bias: Many AI solutions rely on data and algorithms. If the data is flawed or the algorithms are based on false assumptions, then the solution/product will fail. You can avoid bias by developing an accurate understanding of end-user processes and workflows, and by frequently evaluating data and customer feedback.