Month: May 2024

Unethical? Exploitive? Really?

I see numerous freelance translators accusing translation agencies of being exploitive and unethical, often based on their offering low rates for either translation or—more likely these days—post-editing of AI output. I wonder if these freelancers aren’t just embracing a mistaken victim mentality as an extension of their mistaken employee mentality with regard to their position in the scheme of things.

By the very nature of their position in the translation food chain, freelance translators are not employees of translation agencies (most of which are nothing more than brokers and usually have few or no translator employees, but that is a topic for another time). There is no minimum wage or rate. Rather, freelancers are independent contractors who sell translations to agencies that purchase those translations for resale. The rate is decided upon by negotiations between the seller and the purchaser.

As such, freelancers are in no position to assert unethical or exploitive behavior when an agency offers them a rate that does not satisfy them. Translation is a business, and entities engaged in that business (including translators) are free to make and accept or decline offers.

There are, of course, things that an agency can do that would be fairly characterized as unethical. Two are:

  • not conforming to the payment terms (rate and timing) agreed to with a translator beforehand, and
  • getting a freelance translator’s agreement regarding a rate for post-editing of translation done by a professional translator and then giving the freelancer output from an AI system, but still demanding that the work be done at the rate agreed to based on misrepresentation of the nature of the work required.

There are surely other things that an agency could do to land themselves in the category of unethical and even exploitive. But offering low rates—even ridiculously low rates—is not one of them. Translators are free to say no and walk away.

Regarding exploitive behavior, since freelance translators are free agents, there is no basis for the accusation of exploitive behavior because of low rates. Freelancers are perfectly able to say no to a rate, offer other terms, or walk away. I don’t know any translators who are being held in a dungeon by an agency and forced to work for them.

Freelance translators’ complaints that agencies are unethical and exploitive bring to mind some situations in which people with more business sense do not make accusations of unethical or exploitive behavior.

Some examples that come to mind are:

  • when there is a labor shortage, companies do not complain that potential hires are exploitive and unethical by holding out for higher wages;
  • when there is a glut of labor, very few people are so entitled that they are able to suspend belief in the principle of supply and demand;
  • when Betamax format lost out to VHS, Sony didn’t accuse people who no longer wanted to buy their products of unethical behavior;
  • when streaming had its inevitable effect on the sales and rental of physical storage media, companies who made such physical media surely weren’t crying foul; and (a trivial example just for good measure)
  • when the Internet made phone calls nearly free, the Iranians selling unlawfully modified telephone cards on street corners in Tokyo were put out of business, but most appeared to have simply gone into other fields of endeavor.

The act of offering low rates for translation or post-editing is neither exploitive nor unethical. It is arguably optimistic, but the ultimate validity of that argument is yet to be tested.

But even assuming the agencies are being optimistic, the only message freelance translators can send is to walk away. Arguing is not going to work. The power dynamics in the translation business don’t offer that as a reasonable option.

My personal belief is that most translation will in a short time be done to a quality and at a price that will satisfy end users, employing a combination of AI and post-editing. That is not to say it will be of the same quality as when it is done end-to-end by a professional translator, but that business model will probably succeed, because of the supply of translators who will work cheaply enough doing post-editing. But for a freelancer, the best approach if the rates offered are too low is to walk away and find other sources of work.

Translators not able to find other sources of translation work are welcome to:

  • curse the technology advancements,
  • curse their own lack of resourcefulness; or
  • curse the personal circumstances that make finding alternative sources of translation work difficult or impossible.

But it would be best not to accuse cheap translation agencies (again, mostly brokers) of unethical or exploitive behavior based on their business decisions. It’s a business, you see, and not realizing that and acting accordingly just makes you look entitled.

What would AI do?

There are specific reasons not to use AI, which I discuss elsewhere, but today I noticed an AI-related phenomenon that I see becoming more common.

I lately see people faced with problems essentially asking the question “What would AI do?”

AI is not the self-proclaimed son of an imaginary sky friend, but it is being touted as a savior that will lead its believers to everlasting happiness, and they won’t have to die to achieve that; it’s right here, if they just use the right prompt.

In a formerly active but now quite inactive Google translation group, someone cited 整理 as being a troublesome term that comes up frequently. Little context was given beyond one fragmented usage example, and what followed speaks of the realities of freelance translating, while strongly hinting at where we are headed unless we are careful.

Some members of the group immediately launched into experiments with various AI, including ChatGPT, to see what AI would make of the term 整理 (and the somewhat-related term 整備). To be fair, one person also cited dictionary entries made by humans, but again, the linkage to the specific context in this case was missing.

As it turns out, AI of various flavors and origins creates a blinding variety of renderings of the term in question.

Those questions to AI resulted in several pages of different responses. Most are surely wrong for the single cited example of usage, and, in deciding which, if any, might be correct, the missing elements are, of course, the fuller context and the meaning intended by the author, who is (presumably) a human with a context and an intended meaning in mind.

But AI cannot be bothered, beyond spitting out as many varieties of translations as it’s asked for, each with the same Dunning-Kruger level of confidence.

I commented that, rather than experimenting to see what AI would do, the author or someone who can ask the author should be asked just what the context and meaning are. I expect that this comment might be slightly annoying, because it highlights the situation that the translation business is structured such that the task of acquiring this information is not only difficult or impossible for most translators, it is seen by the agencies that feed most translators their work as a costly and troublesome activity. Many simply don’t want to be bothered.

The advent of AI has been a savior to people not wanting to have to care about context, because AI can take over the chore of not caring. No need for us humans to be negligent; an AI can be negligent on our behalf and does it quite well, with no feelings of guilt, before or after the fact.

I thought (for about 3.7 nanoseconds) of asking an AI about the above post, but I couldn’t bring myself to be so negligent or to be so trusting of the capabilities of a pile of computer commands.

A Not-so-Bold Prediction

Well within three years from now, most NES (native English-speaking) translators currently making their living translating Japanese-to-English will no longer be making their living by translating. And they will very likely not be doing post-editing of AI output either, as post-editing rates have already fallen to a level incapable of supporting even a modest lifestyle. What’s more, post-editing is also already being seriously discussed as a target for AI to at least partially handle, and that will lower the level of skills required of post-editors and, by extension, the rates earnable from post-editing.

By “most,” I envision no more than 10% of current NES translators remaining in translation three years from now to the extent that they can earn their living from translation.

Some of the people exiting translation might be able to find employment or work making use of their Japanese-language ability. Some have already done just that. Others will migrate to work unrelated to translation or language skills.

This is not a very bold prediction, as we can already see people leaving JA-EN translation, including highly skilled translators who expected to translate for much longer. It is ending, and people are leaving.

Suggestions that translators simply adopt AI themselves to allow them to continue working are delusionary and ignore the realities (1) that translation is a business and (2) that the businesses (translation brokers) that supply translation work to the vast majority of translators are already switching to AI for the translation process itself, this being a business decision that complaining translators will not be able to change. It is ending.

There are people and organizations to whom the above prediction or even discussion of its possibility is anathema. I would hope that they continue to observe where the profession is going as we move toward the end. It should come into focus for even the most optimistic observers.