AI can’t solve wicked problems A tame problem is clearly defined with a knowable solution; a wicked problem changes with context, involves competing constraints, and has no single objectively correct answer. Infill architecture is a wicked problem. And when Jeff Bezos comes along and says permits should be instantaneous, there is a clear failure to differentiate between tame and wicked problems. So I’m not too worried about being replaced by AI as long as people are involved in making decisions. Because people make a mess. AI doesn’t know what happens when a 70-year-old sewer lateral, a new state housing law, a neighborhood design guideline, and an angry planner who’s wife yelled at her this morning all collide on a 3,200-square-foot lot. It’s computation not judgment. And as long as cities are made by people, infill development will require both.