I find it interesting that AI researchers don't use concepts very often in their theorizing. No doubt they feel no pressure to. This is because most AI researchers do use representations which allow a system to chunk up its environment, and basically all we know about concepts is that they are representations which allow a system to chunk up its environment.