Posted by: patenttranslator | March 7, 2022

Did Machine Translation Finally Killed Translators Like Myself?

Did Machine Translation Finally Killed Translators Like Myself As Was Gleefully Predicted by Some Commenters Already some 15 Years Ago on My Blog?

I thought that it might do that soon, I really did, the arguments of MT enthusiasts seemed convincing, although I tried to counter them with my defense of, basically common sense, or human brain. Common sense was telling me that machines will never be able to think the way humans and translators in particular must be capable of doing to do their job properly. But to be on the safe side, I did file for early retirement when I was 64.5 instead of waiting until the “full retirement age”, which back then was 66.5 in United States.

Although the penalty for doing so was about 12 percent of the actual pension amount (if I remember correctly), I thought it was worth it, as the amount of work coming to me became a relative trickle after 2016 compared to the feast in previous years. I had to be realistic about my future.

I am now into the sixth year of my retirement period, and because my plan for a “post-working” period was very realistic, I don’t need to work at all as all my expenses are covered quite nicely by my pensions. This is partly because I am no longer married, my children are independent, and I live in a less expensive part of the world, compared to California or Virginia where I lived and worked for most of my life.

But although I work much less now that I am officially in retirement, I am still translating German and Japanese patents now in 2022, which is what I started doing when I lived in a small apartment on California and Seventh Avenue in San Francisco in 1987 …. or more than 35 years ago.

The fact is that although machine translation is incredibly sophisticated compared to what it was 30, 20, or 10 years ago, it is still not translation. It is still only a translation tool, a tool that to my old clients – and I only work now for old clients who have been sending me work for more than 20 years – is really not all that useful on its own. Although machine translation (MT), or artificial intelligence (AI) or whatever else we want to call is unrecognizable compared to the situation 35 years ago, it is still only a translation tool that DOES NOT PROVIDE AN ACTUAL TRANSLATION.

And I am still arguing, as I was 35 years ago, that it will never be able to replace human translators. Well, never say never, since I don’t know what may happen in future, but definitely not in my lifetime, which to me is the same as never.

I don’t know what happened in the meantime to the numerous schemes hatched by entrepreneurial translation agencies and would-be agencies that planned to use “post-editors” to massage machine translation into something that would more closely resemble human translation because I stopped following this issue more than a decade ago. My guess is that most of these outfits soaked up a lot of capital from new investors … only to go bankrupt after a while. This is not a way forward, as I was saying all those years ago because real “editing” would be even more time- and knowledge-consuming than an actual translation from scratch.

One thing that strikes me in the machine translation age on an emotional level is how this so-called artificial intelligence is a perfect fit for our new world in which truth, reality and real information, as opposed to vile propaganda, is no longer appreciated or even preferred. As a modern journalist or expert, you can lie as much as you want, it makes absolutely no difference how much actual truth or actual information is contained in your bunch of lies and how far actual information is stretched to fit the narrative. When the winds shift so that what you were saying yesterday is no longer tenable, you can just shift your explanations and instead of recanting yesterday’s truth, or lies to be more precise, you can just say that yesterday’s lies were marketing and that “the science has changed”.

Machine translation offers a perfect tool for an easy manufacture of ever changing shades of narratives. Just add in another mix of well sounding lies, mix it up further with a few semi-truths and then use a cool algorithm to arrive at a predetermined result.

The result may be very different from what you were saying yesterday, but hey, most people are so dumb, they won’t even notice, so as long as your bills get paid with the new, officially accepted truths an d algorithm, who cares.

Although my own example may show that human translators such as myself have not been killed off yet by machines with algorithms, It may be better to use machine translation for processing of information in our age than an actual human brain. It is definitely cheaper and there is no need to feel guilty about the resulting untruths and lies, when lies can be simply defined out of existence and called for instance yesterday’s marketing.


Responses

  1. Steve, I am still working as a technical translator and don’t see any danger from MT. In fact, MT helps me save a lot of time figuring out what the sentences are trying to convey. So, you may say, post-editing is a bad thing, but it does help me say quite a lot of time and they don’t deduct the MT from my rate, so that I can still make a reasonable income. I was for MT and then against it, and now I see no sense to be against it. We must get used to it and make use of it.

    Liked by 2 people

  2. I believe that there is a joke in the industry which says that MT has been about to take over the industry in 5 years for the past 25 years? I first saw an MT system from Fujitsu in the early 90s. They have improved in some ways but gotten worse in others.

    I don’t know about other language pairs, but for Japanese-English, MT can be effectively used in cases where the input is carefully controlled and does not vary much such as a parts manual. However, if you give it even a reasonably complex Japanese source text, it can quickly fall apart particularly when subject information is explicitly dropped. The engine ends up spitting out a vague mashup of pronouns.

    A few years ago, I had a some clients that started to incorporate MT into their work. I made it clear that I was not interested in doing post-editing for a half rate. They argued that it was a “time-saving” tool. However, I explained to them that it actually wasn’t. It was simply much faster to translate it from scratch instead of doing it in my head, looking at what the MT system created, figuring out where it was broken, deciding how to fix it, and then rewriting the “translation.” MT systems often create output with errors that human translators would never commit. The output is really quite alien and tiring to read. Anyway, fast forward a few years and some of these same clients are contacting me about more work after running into too many issues with these MT “solutions.”

    Liked by 2 people

  3. Dear Steve,
    You are very lucky to be retired.
    I work with some of the big names who invest in MT combined with AI and big data. Well, the statement is quite simple for me, I am getting out of translation, these people are going to take over the business. I have tested their systems on very exotic pairs, including patents translations, and they are so efficient that I have decided to get out.
    It is a pity, I loved translation, but there is no future in the business.
    Since I am only 60 years old, I am now trading in cryptocurrencies. I don’t want to get into details, but I make in one day more than in a month of translating.
    Good luck to you and kind regards,
    Ivan

    Liked by 3 people

  4. In my opinion, this is the most important part of your post:

    “One thing that strikes me in the machine translation age on an emotional level is how this so-called artificial intelligence is a perfect fit for our new world in which truth, reality and real information, as opposed to vile propaganda, is no longer appreciated or even preferred. As a modern journalist or expert, you can lie as much as you want, it makes absolutely no difference how much actual truth or actual information is contained in your bunch of lies and how far actual information is stretched to fit the narrative. When the winds shift so that what you were saying yesterday is no longer tenable, you can just shift your explanations and instead of recanting yesterday’s truth, or lies to be more precise, you can just say that yesterday’s lies were marketing and that “the science has changed”.

    Machine translation offers a perfect tool for an easy manufacture of ever changing shades of narratives. Just add in another mix of well sounding lies, mix it up further with a few semi-truths and then use a cool algorithm to arrive at a predetermined result.

    The result may be very different from what you were saying yesterday, but hey, most people are so dumb, they won’t even notice, so as long as your bills get paid with the new, officially accepted truths an d algorithm, who cares.”

    I think that the they will come when they will actually forbid people from learning foreign languages, or will follow Pol Pot’s lead and shoot those who already speak a language other than their mother tongue. The idea is to keep people in cages, and language lends itself to that purpose.

    The existence of someone who can tell that a “machine memory holed” document does not say what the bosses say is going to be a danger for them.

    Liked by 1 person

    • With regards to forbidding people from learning foreign languages. I would say this plan is already well under way given what is left of foreign language teaching in schools and universities…

      Like

  5. There are multiple MT engines out there, and they may give rather different outputs for the same source, just like Linguee or a dictionary gives multiple meanings and translations for the very same word or phrase.
    And who can tell the difference between these variations and choose the best one in context and tone?
    This is a translator! (What was to be demonstrated)

    Like

  6. Yet machine translation cannot be as good as human. Translation agencies use all available CATs but still they require proofreading and editing to receive quality. Working with a text we cannot rely on machine as machine uses algorithms which are far not that ideal.

    Like

  7. No! But let’s be honest, they are getting closer to killing translators like ourselves every year. Up until two or three years ago I was convinced that no system could ever translate as well as a native speaker. Now I’m certain that day will come, and faster than we think. I don’t see a realistic way out for anyone who wants to make his or her living as a translator in the future. Enjoy it while you can!

    Liked by 1 person


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Categories

%d bloggers like this: