There have been some pretty big steps taken in AI research in the last few decades. With the advent of the web, the need to process queries in various languages has natural langauge processing booming, and natural language translation is rapidly maturing. Speech recognition is already a reality, and Carnegie Mellon University has a neural network system that drove a van from Pittsburgh to California, safely and successfully. It's long been known in the medical profession that expert systems are more accurate than doctors at making certain decisions, and that a doctor assisted by an expert system is far more effective than a doctor alone, esp. under stress and pressure.

So we have no HAL and no Data and Deep Blue can't write Sonnets? Please don't ignore the cross-disciplinary work that's taking place in AI right now because of that.

If you asked a robot engineer to build an information processing system out of carbon, she'd say "Carbon? Why carbon? Silicon is much better suited to that task."