I was recently at a webinar about a next generation learning environment, epsilen.com. I’m always excited about the potential of next generation ideas, the built-stuff that follows them is usually disappointing.
Learning objects sounded like it had potential in 2002 when I first started looking into it. When SCORM hit the pavement running, I became uninterested, not only because the US Department of Defense created the standard, but that no one seemed to mind. I still like the idea of learning objects; I don’t think they’ll be used by most instructors in higher ed, at least not in the near future. The term is too hyper-modern, the notion too abstract for most faculty.
The same is true of the term and notion of an intelligent agent.
By and large, I ignore the intellience in Amazon. And I recently noticed the intelligence in emusic; I’ve already started ignoring it too. I may actually see if I can disable it; I spend enough time there to justify the effort.

Intelligent agents are one of those cool ideas. The ones I’m aware of, like the Paperclip in Word, Amazon and eMusic-like vendors who remember what you’ve done, are more annoying than not. They get in the way of my experience. They structure; they categorize; they further dehumanize by acting as if they “personalize” an already impersonal experience.
Vendors are about commerce and there are rules of commerce. Since I don’t follow them for the most part, the agents don’t work.

I use help menus extensively now and have acquired a certain intelligence about them. If I’m stuck, it’s often because I don’t have the language to find the answer. The paperclip can’t give it to me either.

The missing links of intelligent agency are linguistically and contextually bound, utterances and contexts are infinitely unique and profoundly human.