He wrote about this some on his blog: http://blog.asmartbear.com/vetting-startup-ideas.html

I don't see why AdWords couldn't work if you built the right experiment. For example, a simple email-capturing landing page with a price tag doesn't touch on it as much as personally talking to people and finding 30 who will (not would, as Jason emphasizes) pay at that price. You can learn a lot from just talking to folks, and an AdWords campaign could help you get in touch with the right people.