Monday, June 10, 2013

Mathematical language

Recent research from the University of Texas demonstrates how Markov logic chains could be used to assign a probability of “closeness” of a word or phrase to a paraphrase. This appears to finally break the shackles of determinism and prescriptive rules in natural language processing. Decades of lull in artificial intelligence could be attributed to those trying to ram a square peg into a round hole. Chess and Jeopardy have given certain technologies respect they do not deserve. More importantly, research in that direction holds everything back. It is unclear if computer scientists will ever accept that intelligence is not rules based.

Intelligence, however, is not rules based. Engineers with traditional training find it hard to move away from this concept just as the business students who were given antiquated methods to determine the value of something, fail to consider alternatives. This is a significant failure of the education system that appears to favor the past and discount the future. Why else would we have people in venerable companies stuffing rules into hardware in an attempt to beat a human or business students who adorn spreadsheets with numbers neatly adding up, down and sideways as if it means something. Language is a good example of chaotic outcomes that start with simple patterns. Incremental addition of random noise renders it completely incomprehensible to those with linear brain waves.

If the education systems are not able to catch up with the present, we may be fast approaching a regime in which education has negative value.