This post originated from an RSS feed registered with .NET Buzz by Eric Gunnerson. Original Post: Intelligent Agent Nirvana Feed Title: Eric Gunnerson's C# Compendium Feed URL: /msdnerror.htm?aspxerrorpath=/ericgu/Rss.aspx Feed Description: Eric comments on C#, programming and dotnet in general, and the aerodynamic characteristics of the red-nosed flying squirrel of the Lesser Antilles Latest .NET Buzz Posts Latest .NET Buzz Posts by Eric Gunnerson Latest Posts From Eric Gunnerson's C# Compendium

At the dinner last week, we got into an interesting discussion about intelligent agents.

The discussion is around whether intelligent agents will be able to gather the information that we want to see automatically, or whether human intervention is necessary.

I should probably note here that I'm simplifying the discussion, to keep things short and to make me look better.

My assertion is that intelligent agents not only aren't there yet, but are unlikely to be there (wherever "there" may be) in the forseeable future.  Now, in making that pronouncement, I am aware that the track record of people saying that thing are impossible - such as the crazy notion of "heavier than air" flight - isn't exactly stellar.

So, why don't I think that intelligent agents are going to work - at least for me? Well, a few reasons.

The first is my skepticism around anything that requires AI. Way back in the early 80s, there was lots of press around AI systems, with a ton of money being spent both by DARPA and the Japanese, and no real results. I think that's a good demonstration that "AI is hard", and I don't expect any breakthroughs in that area.

Another challenge is that I really don't know what I want to know. I continue to find offbeat information in blogs that I didn't know that I wanted to know, so I don't see how I can expect an agent to filter in that manner.

Finally - and related to the last point - I think that coming up with a categorization system that works well is likely to be very, very difficult. If you've ever fought with Google trying to find the one page in 100,000 on a specific topic that you looked at a few weeks ago, then you understand what I'm talking about.

I expect that a human filter will remain a necessity to get good information for quite some time.

So, what do you think? Will agents make human editors obsolete?

FYI, here's a brief history of AI, and a wikipedia article