“The pilot is the last line of defence when things go catastrophically wrong. In January 2009 an Airbus A320 hit a flock of geese over New York City. With no power, the captain, Chesley Sullenberger, had to weigh up a number of options and act quickly. Using his extensive flying experience and knowledge of the plane’s handling qualities he elected to ditch the aircraft in the Hudson River. The 150 passengers were not saved by computers or any other automated system. They were saved by the two pilots – the very components that techno-enthusiasts claim can be replaced by computers and ground controllers.”
You see, the claim that humans can be replaced by machines is not made only in the T&I industry. I recently went to Panera Bread (a fast-food chain, though much more healthful than McDonald's) and placed my order on a tablet that was docked not too far from the entrance. The sign by the tablets claimed that this would expedite my order. A little further inside were actual people taking orders behind a counter, for those customers unable or unwilling to use the tablet. I felt guilty for using the tablet and promised myself to place my order with a person next time, even if that means having to wait in line. I refuse to use a machine and contribute to making a person lose his/her job just to save a couple of minutes.
Last year on a trip to Boston I had to rent a car at a very late hour because my flight had arrived late. Most rental car desks at the airport were closed. I saw that in a couple of them there were machines which you could use to rent a car, without the help of a person. In fact I saw something similar on TV a few weeks ago, in a commercial claiming that now you can rent a car even if you don't feel like talking to a person. So not only was that ad discouraging human contact (all right, maybe one is anti-social and really doesn't want to talk to anyone; fine) but it was also suggesting that it's OK to replace a human being with a machine.
However, when that machine breaks, who are you going to call? You are going to call the specialists, the ones that know how to rent you a car, the ones that know how to land a plane even when things go wrong, and the ones that can recognize immediately why a machine translation may be nonsensical and even deadly (numerous are the examples we’ve seen of bad medical translations produced by a machine, such as incorrectly translated directions for taking medication; and don’t even get me started on how many translation errors I’ve corrected in aircraft specs and assembly instructions).
So those that still think they can save money with machine translation should probably do a more detailed cost analysis and take into account the cost of fixing what will be broken. Because chances are, something—or many things—will be broken. One of the most important things that will be broken—and that will be really difficult to fix—is the reader’s/user’s trust. The other day I was reading an article in Deleátur (the journal published by the Union of Spanish Editors) on the importance of proofreaders and editors and on the negative impact of language errors found on websites (http://www.uniondecorrectores.org/img/web/docs/deleatur_0.pdf). The article referenced another article, published on the BBC website, which claimed that bad spelling on English websites costs the UK millions of pounds and that errors on websites can reduce online sales by up to 50% (the author went on to explain how this was measured). So we are making people all the more redundant, and for what? For the better? Are we really saving money or are we wasting more money and time? Because when I saw that the order wizard on that tablet at Panera Bread did not ask me whether I wanted chips or bread or an apple with my sandwich (you have to pick one of the three to get with your order), I realized that I had to go talk to a person, for which I’d have to wait in line, which of course would defeat the purpose of the tablet. And I was afraid to use the machine to rent a car at the airport in Boston because what would happen if it charged my credit card but didn’t give me a key? So I chose to go to the only desk that was open at that hour and where someone capable of a meaningful dialogue was present (well, as meaningful as a dialogue can be at 2 a.m.).
But are humans useful only for debugging and fixing what’s broken, used only as a safety measure? Numerous articles have been written on why humans cannot be replaced; many of them by translators, explaining why translations should be done by humans and giving innumerable examples of bad translations and the damage that they have caused companies and individuals. Many articles have also been written by business people claiming that a company can save money by using machine translation. In fact I recently read one on the IAPTI (International Association of Professional Translators and Interpreters) LinkedIn group. Think about it: somebody who isn’t a translator published a blog post in a translators’ group, claiming that it’s a good idea to use machine translation instead of a real translator. It’s not the first time I see this. In fact a few years ago in one of my talks at a translators’ conference I talked about “infiltrators”, who come to our conferences and our online groups and try to sell us the idea that machines are better/more productive/more cost-effective for translation, that machines are preferable to us translators, and that the smart thing for us to do is embrace post-editing because allegedly it’s the future and it’s inevitable. The nerve!