Will Machine Translations Be Better Than Translations by a Human Professional?
A complete take-over of the human race has been predicted by some for a long time. It has not happened yet, nor does it look very likely in near future. One theoretical limit is that unless artificially created intelligence can exceed that of human beings, the takeover cannot happen. Unless this happens, and the robots equipped with the artificial intelligence learns what human beings know in its totality, the translation problem is also not going to be won by the machines.
For complete and accurate translation, not only the machine translator has to be knowledgeable in the pair of languages involved. They need to know a lot about the people who use the two languages. Human translation professionals not only know the languages, they know the phraseology, the context of the use of languages, the various nuances of each word, etc. The words with shades of meaning and the exact use of context, any special use and the historical and cultural backgrounds of the two language user groups are also known to them. It is with this kind of specialized knowledge; it is possible for professional translators to do a complete job of the translation.
It is this large amount of peripheral knowledge. That is essential for accurate translation. It is also necessary for the translator to know even the exotic languages when it is one of the languages involved in the translation project. Unless, machines doing the translation become knowledgeable to an equivalent level, it is impossible to become as capable as the professionals. Artificial intelligence, learning capabilities are to be developed by researchers to those levels before machine translators can reach and exceed the levels of the human professionals.
If one were to look at the developments in AI, it has not really delivered on the promise. There are examples of the use of AI in very limited domain areas, but it has taken more time than predicted. Intelligence levels and the learning capability of a human child is far from being equalled yet. It is difficult to foresee a time when a machine would exceed the general capability of human beings. A specific machine beating a chess grandmaster or a machine beating the top human player in "Jeopardy" is exciting news. However, do they represent a general progress towards human intelligence being overtaken; hardly.
Natural language processing is the mainstay of the building of the translation capability in machines and has a lot of hurdles to be crossed. It will be difficult to predict how long it will take to surmount those problems. One would then need a combined processing capability of additional skills brought to use by human translators. These extra capabilities would be the intimate knowledge about use of the languages and the nuances of meanings of words as used by the actual users. Equipped with the right algorithms and armed with all this contextual knowledge will the machines be able to equal (may be exceeded) human capabilities!
Debasis currently holds the chair of two departments, Computer Science and Information Technology departments in a 4 year Engineering & Technology college. He also consults on Embedded Systems, Software development & Testing, coaches students of Electronics & Computer Science, and manages content development. He has worked with technology companies for close to four decades, 25 years of which has been with IT consulting companies. He has managed software and Geospatial industry outsourcing from India for International clientele. He has worked with customers from US, Europe, Japan and China. He is widely traveled.
For complete and accurate translation, not only the machine translator has to be knowledgeable in the pair of languages involved. They need to know a lot about the people who use the two languages. Human translation professionals not only know the languages, they know the phraseology, the context of the use of languages, the various nuances of each word, etc. The words with shades of meaning and the exact use of context, any special use and the historical and cultural backgrounds of the two language user groups are also known to them. It is with this kind of specialized knowledge; it is possible for professional translators to do a complete job of the translation.
It is this large amount of peripheral knowledge. That is essential for accurate translation. It is also necessary for the translator to know even the exotic languages when it is one of the languages involved in the translation project. Unless, machines doing the translation become knowledgeable to an equivalent level, it is impossible to become as capable as the professionals. Artificial intelligence, learning capabilities are to be developed by researchers to those levels before machine translators can reach and exceed the levels of the human professionals.
If one were to look at the developments in AI, it has not really delivered on the promise. There are examples of the use of AI in very limited domain areas, but it has taken more time than predicted. Intelligence levels and the learning capability of a human child is far from being equalled yet. It is difficult to foresee a time when a machine would exceed the general capability of human beings. A specific machine beating a chess grandmaster or a machine beating the top human player in "Jeopardy" is exciting news. However, do they represent a general progress towards human intelligence being overtaken; hardly.
Natural language processing is the mainstay of the building of the translation capability in machines and has a lot of hurdles to be crossed. It will be difficult to predict how long it will take to surmount those problems. One would then need a combined processing capability of additional skills brought to use by human translators. These extra capabilities would be the intimate knowledge about use of the languages and the nuances of meanings of words as used by the actual users. Equipped with the right algorithms and armed with all this contextual knowledge will the machines be able to equal (may be exceeded) human capabilities!
Debasis currently holds the chair of two departments, Computer Science and Information Technology departments in a 4 year Engineering & Technology college. He also consults on Embedded Systems, Software development & Testing, coaches students of Electronics & Computer Science, and manages content development. He has worked with technology companies for close to four decades, 25 years of which has been with IT consulting companies. He has managed software and Geospatial industry outsourcing from India for International clientele. He has worked with customers from US, Europe, Japan and China. He is widely traveled.
0 commentaires