Artificial intelligence has always been a field which goes though wild swings. People either pin lots of (unreasonable) expectations on it or blame it for (unsupportable) hype. The fact remains that AI is an intellectual discupline that addresses some challenging problems. Like any other field, there is good research in it and there is bad research. What is different about AI, compared to other disciplines, is the fact that it addresses issues related to human intelligence. Since human intelligence is one of the least understood but the most important human feature that has made us what we are in the ‘live’ world; people get excited about it.
DARPA recently conducted a grand challenge race for autonomous vehicles in the desert of Nevads. This has been attracting lots of headlines and people have been calling it a test of AI. It does have AI component, but it is equally strong engineering challenge. In any case, the race is over and it was more successful than similar previous races. For more details, see for example, an article in NYT.