Read it here.CUTTING-EDGE artificial intelligence it was not, but a student prank still managed to get the better of some human intelligences last week, when a computer-generated piece of gibberish was accepted as a genuine scientific paper.
Sick of receiving spam emails requesting submissions to the 2005 World Multi-Conference on Systemics, Cybernetics and Informatics - which charges $390 for each attendee - students Jeremy Stribling, Daniel Aguayo and Maxwell Krohn of the Massachusetts Institute of Technology wrote a program to generate a nonsense paper.
Starting with skeleton sentences, pools of nouns, verbs, adjectives and adverbs, and a random assortment of computer science jargon, the program produced a grammatically correct yet utterly nonsensical paper titled: "Rooter: a methodology for the typical unification of access points and redundancy". "This isn't artificial intelligence, it's the dirt-simplest way we could think to do this," Stribling says.
The conference organisers say that the paper was sent to human reviewers, who never commented on it, so it ended up being automatically accepted. The conference has now banned the paper. But the pranksters are still planning to give a computer-generated talk at the conference by persuading a human speaker to let them take his or her place.
This is more than just a student prank. There is a serious point to be made here.
The entire authority of the scientific establishment [indeed, of any intellectual community] rests on an assumption that the practitioners are competent and disinterested seekers of truth, whose competence is certified by accrediting and professional institutions.
This is just one in a long string of recent instances in which those certifying institutions have been shown to be incompetent, corrupt, or biased. If we cannot trust the certifying institutions, we cannot trust the practitioners.
No comments:
Post a Comment