Friday 18 November 2011

The Google Translator: testament to a human miracle




When feats of artificial intelligence are reported in the media it is easy to be deluded into feeling that the machine in question has some intrinsic merit and forget that what is remarkable is the human input.  So when we learn of some new peak of computer engineering we need to keep in mind the creativity, ingenuity, intelligence and perseverance behind it.

For instance, the Google Translator.  One can enter a phrase, even a colloquial one, in, say,  Russian or English and the English or Russian equivalent comes out. The same applies to numerous other pairs of languages – even fundamentally different ones like Chinese(?) and English. Poetry, or long passages of prose, cannot of course be made sense of. Nevertheless, it is a major achievement.

Several decades ago machine translation was almost useless and only incremental improvements were made over the years.  According to an article in the New Scientist (Oct 15) the big breakthrough was in the early 2000s, when Pentagon researchers started to programme computers to search for language patterns rather than individual words. Vast digitised collections of documents, such as millions of UN conference transcripts in six different languages, have enabled powerful pattern recognising algorithms to be developed. Google took over the research after hiring Franz Josef Och who had won a machine translation competition in 2003.

It is difficult to know how far back to go to appreciate this achievement. One could start with the dawn of creative thought, perhaps ten thousand years ago. That would be beyond the scope of a post and take too much time, so let’s restrict ourselves to developments since the 19th century. All the following and probably more, had to occur before the software of machine translation could be written (no need to read the whole list – it’s just to give an idea of the scope of work which led to computer science ).

  • Mathematical developments, including imaginary numbers which are of particular value in the theory of electronic circuits.
  • Laws of electromagnetism discovered and expressed in mathematical form made electrical engineering possible.
  • Major advances in the science of chemistry, including the arrangement of elements into the Periodic Table.
  • Semantic analysis of languages around the world.
  • Invention of the electronic valve, capacitor and other electronic components.
  • Invention of the first punched card computers using valves.
  • Theoretical and experimental investigation of quantum theory and semiconductor band structure.
  • Invention of the transistor.
  • Painstaking characterisation of silicon and other semiconductors by research teams around the globe.
  • Collation of peer reviewed research findings into a coherent body of knowledge on semiconductors.
  • Invention of innumerable techniques in chemistry, crystal growth, purification, implanting of elements, cooling, heat processing etc.
  • Invention of the integrated circuit and the silicon chip using this knowledge.
  • Invention of the first computers based on chips.
  • Advances in information, circuit and computer processing theory.
  • Advances in pattern recognition.
  • Continual improvements in the performance computer chips (Moore’s Law etc.).
  • Continual improvements in programming languages.
  • Availability of large numbers of documents translated by people into a large number of languages.
When a new supercomputer is announced, or a robot, or new medical device, it's easy to forget that  it embodies in the present miraculous chains and networks of human inspiration and hard work over a long period. That is what is remarkable, not the technology per se. It all springs from the unique and precious qualities of the human being, which, I believe, reflects in turn the image of our Creator (Genesis 1:26).

John
Author, 2077 AD (being revised)

cosmik.jo@gmail.com