When Cyc started there were a lot of successful rule-based expert systems that the AI community was really exciting about, doing things like medical diagnosis and credit card fraud detection. It's a very good way to automate making the sort of routine decisions experts in a particular domain make within that domain. As weak AI, that model is excellent for modeling what it models. Using the informal definition that an intelligent program is one that does what an intelligent person would do, I have no problem calling those programs intelligent.

The techniques we use in AI are all just magic tricks though. They're really cool and really useful magic tricks, but they don't tell us anything about our own minds, or about how to write programs that are intelligent in the way we are. Maybe they will, eventually, but we're far from that point.

I don't think any of these guys are dumb, they've all done clever work, they're just making assertions way beyond what they're justified in asserting as scientists. When people do that with quantum physics or Gödel we call them cranks.