I became a die-hard advocate of on-demand apps since they started cropping up about 5 years ago. It’s made getting a ride easier, ordering take out, and even having the curtains hung up–all thanks to on-demand app services. However, a couple of months ago I had a change of heart about everything “on-demand” when it came to machine translation. It started when a long-time client, who I had not heard from in a long while, sent me a message. She asked for me to be on a team of translators for a special corporate project. The subject matter was unusual: military memorabilia, fine watches, antique coins, diving equipment and sex toys! Subject matter that demanded a nuanced understanding and vocabulary for magazine copy. I was surprised and excited at first. Then I was shocked when I read on my screen the tiny black and white decimal point mocking me as I looked at the dismally low per word rate. Why? What? So, I read on.
My job on the special project was post-editing machine translation. I was to correct the machine translation, and then the engineers on the other end would feed it through the machine. So, this is what is called “training” the machine model. The rationale behind the low rate was it had already been translated once. I had to correct it. Then the machine translation would be perfect.
For the first time, I think I felt like so many factory workers who lose their jobs to robots. One day you walk onto the factory floor and there it is: a big fast elegant looking machine swinging about. It never gets tired, rarely makes mistakes, and can do the work of 100 men and women. And most importantly you only pay for it once. “Are translators being downsized?”, I thought. It’s a serious question now. 5 years ago no, but now it is. Technology increases exponentially.
Obviously, there is no comparison between translation and manufacturing skills. Though I realized that for the “client”, what I produce in translation is not unique. The words are the same as what this or that person would choose. The sentence or phrase is no different than a rearview mirror or maybe a cigarette lighter. Maybe it doesn’t matter with magazine copy, who knows? But there is intrinsic value in the human touch. Isn’t there?
I did the project. I felt a little dirty admittedly. Of course, I did a good job. I did an excellent job. I cannot help doing the best I can. Besides I needed the money—famous last words. It just made me realize that like it or not, the future of translating magazine copy may be eliminated by artificial intelligence. And it’s not a man in a black cape and mask trying to take my job. Machine translation is big business. Just look at TAUS for example. It’s an international organization bent on the destruction of the ‘human-touch’ translator. The conflict is, I understand why they are doing it—why they are collecting the world’s translation memories into a massive database. Smartling is another massive organization that has streamlined technology localization. One plug they had somewhere mentioned that the Internet has so much text it is virtually impossible for all the world’s translators to come together to take on the task of translation. I cannot disagree with that.
So how do I weigh the benefits against my checkbook? In my opinion, one of the greatest detriments to social advancement is the language barrier. If language were not an issue, is it possible we could soften war and conflict, or at least make it a lot easier to get to know our earthly neighbors? Let’s be honest though. No corporate entity is genuinely egalitarian. It’s not fundamentally possible. So, here we stand at the edge of a social paradox: on the one hand machines, or machine translation as the case may be, is supposed to speed up and provide a generally high-level product. On the other hand, we are violating key economic principles which dictate that a healthy division of labor results in the creation and accumulation of wealth. What is to happen to the translator? Are we to become old out-dated machinery?