Before asking AI for a solution, we need to examine what we are describing as the problem. Is what we see a symptom? What outcome needs to change? AI can generate many solutions to a well-defined problem in seconds, but it can amplify a poorly framed problem just as quickly.
When trying to understand a problem, I make a point of using more than one method. A fishbone diagram helps examine possible causes under headings such as people, process, technology, environment, and management. The Five Whys moves us toward the root cause. Socratic questioning brings assumptions, evidence, and counterexamples to light, while TRIZ places contradictions at the center of solution design.
During this investigation, AI can group observations from the team, suggest causes that may have been overlooked, and generate contradictions and alternative problem statements. We can also ask for missing questions, alternative problem frames, and counterevidence before requesting solutions. People decide which data is reliable, which causes are meaningful, and which problems are worth solving.
It helps to rewrite the problem statement several times before moving to a solution. Clearly describing the person affected, the unwanted outcome, the context, the evidence, and the boundaries clarifies what we are trying to solve. Removing solution language from the initial definition also matters. This lets us examine causes and contradictions without committing to a particular solution too early.
Ask yourself
- Is what we are trying to solve the problem, or merely the visible symptom?
- Which assumption are we treating as fact?
- How does the problem look through a different stakeholder's eyes?
