Translation Losses
Either the horror that is artificial intelligence (AI) is lurking in the background, toying with us all, about to pounce and make the living envy the dead, or else the seeds of a better, brighter human tomorrow are silently sprouting beneath our feet, and we have only begun to grasp the AI promise as we join hands and amble gaily into the brightly lit world of tomorrow. The technology is grinding onward, whether we like it or not, whether we believe it portends cataclysm or unimaginable marvels. Anyone who wants to pause it is merely asking that it become an exclusive tool for the most advantaged elites and those who work entirely outside the laws of society for individual or group advantage. In this telling, there is no middle ground on which to stand and remain unaffected.
It’s going on 15 years now since I was a translator. It is an odd line of work. Translating is a task performed in the background by a lone individual typing as quickly as the words occur in the target language in response to reading words in the source language. The way it was 15 years ago, translating texts from one language to another was predominantly a commercial and marketing activity done for business clients working across international borders. Most translating work was—and still is—done by freelancers: people who waited for jobs to arrive, almost always under terms of the utmost urgency.
In my personal experience as a translator, the job required waiting at the ready for jobs to appear, and then suspending all other activity to bang the words out as quickly as possible. Time also had to be reserved for revision and editing before sending the text on to an intermediary who re-reviewed the final product and returned it to the customer. The perpetual waiting by the computer became a time to waste on the internet, interacting with strangers in online forums talking and arguing over politics, among other things. You can’t wait at the ready to do translations if you leave the desk and interact with live human beings in the non-virtual world. This constant self-isolation was the main reason for me to give up the trade.
Somewhere after I quit the industry, Google introduced a capable translation tool online, based on large language databases. This has only improved over time. Google Translate and similar products now offer immediate translations for a wide range of languages—for free, faster than humans can do, as convenient as speaking into your smartphone. And this, thanks to AI, is what the future looks like in many areas of so-called knowledge work. The algorithm behind translation software is a cousin to the large-language models behind AI text generators like ChatGPT.
Cheap and easy competition from computers, as Tim B. Lee reports, has driven down costs and earnings for translators. At the same time, it appears to have increased demand for translations from human translators in general because of the improved affordability.
The advent of machine translation has not been catastrophic for human translators, but machine translation software does seem to be putting downward pressure on wages. And it’s easy to imagine that pressure intensifying in the coming years as AI technology improves further.
Lee finds a parallel in the realm of taxi services and navigation apps used by rivals like Uber and Lyft.
Before the advent of smartphones, driving a taxi required extensive knowledge of local streets. Indeed, New York once required cabbies to pass a geography test before they could get a license.
Smartphones changed that. The Uber and Lyft apps supplied turn-by-turn directions that allowed unskilled drivers like me to drive without any special training or experience. That was not a positive development for existing taxi drivers because it devalued their hard-won navigational skills. But it was good for consumers, who got more ubiquitous and affordable taxi services.
Translation is a decidedly unglamorous trade, considered to be as tedious as driving a taxi. The most tedious part, however, is often reviewing the final product. This is the main job required in machine-human hybrid translations. Professional translators responded to Lee’s article in the comments to mention how time-consuming the review work is in reality. After all, you have to read and comprehend the source text, the machine translated output, and then try to spot mistakes.
That is to say, this is what the work looks like today. The pace of improvement to AI software will likely make parts of this work less tedious in the future, too, for instance by flagging sections of the text for extra human attention. Nevertheless, productivity for each translator will likely improve, with fewer translators needed to handle increased volumes of professionally translated language. And perpetual complaints about downward pressures on pay will go on as usual.
Good morning. My friendly debate with M. J-C J (about the "light and fluffy" characterization) continued, though it's about wrapped up, I think. The thread is on Friday's G-File, "When Reality Is a Punch in the Face." It went on long enough that some comments don't allow a reply other than by attaching it to a previous comment.
I think this link will take you to somewhere in the middle of it: https://thedispatch.com/newsletter/gfile/when-reality-is-a-punch-in-the-face/?utm_source=reply_notification_email&utm_medium=email#comment-977049
At any rate, whoever's right about "elite good" or "elite insufferable," I believe I have the right to claim he can't out-Lewis me.
I will use Google Translate for quick and dirty purposes, but if I have to be sure of what a word means, and what it means in the context, I translate results back and forth and do more searches just to be sure I haven't missed any unintended connotations of that choice, or overlooked a word that might work better.