You say people should chase simple, realistic, high-demand AI ideas. Why?

Many people hear AI and think, 'I need to build the next ChatGPT.'' That is a very expensive, hard game. I prefer ideas where demand already exists and you can deliver with a small team. For example, helping companies integrate AI, build content, or launch agents. These are not sexy research projects, but they can be seven or eight figure businesses.