ศาลาชุมนุม

Friends,

For those of you with an interest in language and computer systems, I highly recommend an article from the New York Times Magazine titled "The Great AI Awakening, How Google Used Artificial Intelligence to Transform Google Translate . . . "
If you have a chance to read this somewhat long article, please let me know what you think. Thanks.

http://www.nytimes.com/2016/12/14/magaz ... ctionfront

Yes it is long, too long. I like practical, so was interested in the claim that only the indefinite article missing from leopard identified the Google translation. Originally "The Western summit is called Masai "Ngaje Ngai" The House of God. " Three names, that would disqualify the original for me! I think that his translation to Japanese must have taken it to be a typo and changed to: called 'in' Masai which comes out in the back translation. I would be interested in how Google would have understood it without "in".

I tried พวกพ้อง in Google translate and got 'Partisan', singular you will notice.

Tgeezer,

It is not clear to me whether GT's "Thai" has been updated yet or not.

Maybe I should have read more of the piece. I thought that the programme was learning and thus improving itself. A translation needs only to have familiar syntax and the more traffic it gets between two languages the more likely it is to recognise correct syntax. I agree, Thai-English traffic is tiny compared with Japanese=English so it isn't good for us.