Saturday, December 02, 2006

The Human-Computer Interface

Librarians have long understood that "artificial intelligence" is more often an illusion born of the interface between human intelligence and computer technology. Many legal researchers have a faith in the results of computer-assisted searches that borders on assuming it has true artificial intelligence. It seems to be a natural human assumption; I feel it myself. If I have searched repeatedly online, and found nothing, I feel convinced there is nothing there. But it is a false sense of confidence. For instance, I tell my students that whether they use "terms and connectors" searches or "natural language" searches, their results will always depend on their personal success envisioning the document of their dreams. They must use the terms that will be in the actual text in a "terms and connectors" search. They must succinctly state the issue and select appropriate synonyms from the thesaurus in a natural language search.

Last Sunday, there was a wonderful article in the local paper proposing the term "intelligence augmentation" for the use of humans to augment computers by providing the tasks that people do better than machines. Link
here to the Boston Globe's article, "Souls of a New Machine." Examples given in the article are Wikipedia and the new "game" where people go online to donate labels to Google Image Labeler by competing against each other to name more images in a few seconds' time. In both cases, humans are adding to information databases in a way available only when computer networks provide the "architecture of participation." The Globe article is based on a blog entry, see the original here at Tim O'Reilly's Radar blog from March, 2006:

... the old dreams of artificial intelligence were being replaced by this new model, in which we are creating more intelligent systems by using humans as components of the application. Tom neatly summed up the paradigm shift: "AI becomes IA." ("Artificial Intelligence becomes Intelligence Augmentation.")

This is the power of memes: they are framing concepts that help you to see the world in a new way. Now that I understand that we're building a next generation of bionic systems, I'm seeing them everywhere.

The weakness in this "bionic" partnership can often come from the human side of the equation. We are all aware of the edit wars and vandal scandals of Wikipedia. Now, there is a new aspect of those issues on the Chinese version of Wikipedia: Self-Censorship. C-net
carries a story comparing the English version and Chinese version of Wikipedia through their treatment of Mao Zedong. Despite edit-fights in the Chinese version, a carefully self-censored version prevailed, omitting any mention of the famine caused by Mao's Great Leap Forward and the brutal repressions of the Cultural Revolution.

The power of the Chinese government's censorship and control of education extends itself through the information offered by those controlled minds on the Chinese Wikipedia. Once again, the "intelligence augmentation" effectiveness founders on the fallibility of human intelligence. We do so many things easily that are nearly impossible for computers. This partnership, this form of bionics will certainly be continuing in the future through a multitude of projects. But we need to test and re-test the results, to avoid falling victim to our human foibles.

Image is from the Boston Globe

No comments: