Rise of the Machines

I recently read that a noted futurist (Ray Kurzweil) reckoned that machine translation will catch up with human translation in about 2029.

So, does that mean that in eighteen years all human translators will become redundant? Quite frankly, I find it hard to imagine. Certainly, at the moment, with the best will in the world, machine translation generally requires a large overhaul by a human editor to end up looking like a natural specimen of the target language.

Of course, the technology is improving and with immense multilingual corpora (phrase and sentence data collections) and context recognition technology, something like Google Translate can recognise phrases it has seen before and place the appropriate translation into a recognised context but it will still, as often as not, produce something ungrammatical or, worse, nonsensical.

As things stand, the advantage of machines over humans is that they can systematically sort through vast amounts of data very rapidly. But this brings with it inflexibility and they generally cannot deal with completely new text or decide on the best of multiple translations of the same source text.

Another issue is that there is no such thing as the perfect translation, particularly when it comes to more literary texts, puns and other wordplay. How does a computer cope with this?
IT might present options but a human should have the final word. (Now translate that sentence!)

Kurzweil himself does not suggest that human translators will be replaced but states “These technologies don’t replace whole fields, in general. What they do is replace a certain way of applying them.”

All in all, I think it will be a long time before we see the profession of translator terminated.