Is it possible that $712,883 in stimulus funds was used for the development of computer generated humor? Just another little drop in the phony stimulus bucket. No big deal, after all it is less than one tenth of one percent of the total amount of stimulus dollars and in the age of Obama we all desperately need a good laugh.

The researchers at Northwestern University in Evanston, Illinois (just north of Chicago) are designing an artificial intelligence (A.I.) that will be able to do stand-up. Jokes will be gleaned from the internet and integrated into a sure-fire, wiz-bang, idiot proof comedy routine. The avatar comic will deliver punch lines with mechanized precision enhanced with hilarious computer generated expressions and gestures.

The new A.I. comedy program will be based on "News at Seven," which was designed by Nathan Nichols and Kristian Hammond at Northwestern University. The new system represents another evolutionary step in machine generated multimedia.

According to the university's abstract on "Machine-Generated Multimedia Content," "News at Seven" was designed to collect news and information from a variety of diverse sources and then integrate that data into homogeneous internal structures. By incorporating the data with video, animation and computer generated speech, "News at Seven" can produce a full newscast which is delivered by an avatar anchor person. Because the content is machine generated and sent to users on the internet "News at Seven" can take into account viewer preferences and then custom tailor the newscast accordingly.

Although the "News at Seven" program is a remarkable new technology and has been successfully tested, there are a number of system limitations which the designers are working to correct. The avatar anchors deliver the content in a boring monotone and lack the ability to enhance the experience through the use of natural expressions and reactions. Current research on machine generated humor should yield results which could be applied to the "News at Seven" system and allow it to reach its full potential.

Maybe it's just me, but I find these machine generated multimedia projects rather frightening in an Orwellian sort of way. By utilizing a user's preferences (accumulated personal data) to custom tailor the newscast, it appears that the A.I. would have the capability to subtly influence the viewer.

Suppose that a particular user has a visual preference for Chinese women (21-30) and an auditory preference for a British accent, the A.I. could create a user targeted avatar and give her a suitable name and background. The unsuspecting user would have a natural attraction to his custom designed avatar. How long would it be before news and information were formatted to fit a particular political or social agenda and then incorporated into the newscast? Could the system be used as a behavior modification or psychological conditioning tool?

O.K. call me paranoid, this could never really happen here in America. It is possible that this new technology is being developed for TOTUS.


paboehmke@yahoo.com