Google has just created possibly the world’s best computer program designed to understand the English language but this cutting-edge technology may just be the tip of the iceberg.
A new theory could be about to unlock the secrets of how naturally human language really works, putting an end to a question that has troubled linguists for more than half a century.
By empowering artificial intelligence with these secrets, it may finally be possible to create a ‘bolloxometer’, which could instantly cut through the rhetoric of language online to determine its reliability.
No more lying politicians. No more lawyers spinning a story. No more advertisers hoodwinking us into buying their products. Just switch on the bolloxometer and watch the truth unfold.
Google researchers are getting a bit of software widely noticed by calling it Parsey McParseface. They picked the name on a whim after they had failed to come up with one that suited after 18 months of work on the project. It comes after a public vote almost saw a £200 million British polar research vessel named “Boaty McBoatface” after more than 100,000 people voted for the name.
Parsey McParseface has major importance. Parsing a piece of language correctly is the key to getting the computer to truly understand what language means and to be able to act on it.
Up to now, machine parsers have been limited to working out the ‘parts of speech’ of words and their structural interrelationships within a sentence. That’s useful. For example, in speech recognition, the computer can reject analyses of speech sounds that add up to ungrammatical sentences.
But there are problems. The computer cannot easily reject grammatical-but-nonsensical analyses. And even the best parsers struggle with very long sentences and with unusual constructions and idioms.
Actually the statistical methods had to be adopted because linguistics failed to develop a full set of grammar rules from the ideas of Noam Chomsky .