Jump to content
💬 Smartcat Community
Vova

“Are low rates and robots coming to get us?”

Recommended Posts

@Corinne McKay recently wrote about low rates and machine translation (thanks to @Flavia Luz for sharing!).

You can read the whole article here, but I would like to copypaste some quotes that I find especially spot-on.

First of all, Corinne lists some cases where MT is clearly applicable:

Quote

I’m not an MT hater, at all. As I often explain to my clients, I use Google Translate myself…just not for anything other than a general gist of meaning. If I’m going on vacation to Italy and want to know what days a museum is open and closed, Google Translate does the job. When a client in Switzerland forgets that I don’t speak German and e-mails me about a translation auf Deutsch, Google Translate can usually provide a pretty good approximation of what they’re asking.

and some examples more relevant to the translation industry:

Quote

If I translated for industries where translations are mostly done for informational purposes only, mostly include only a limited range of vocabulary, and where there’s some tolerance for bad writing or even minor mistranslations (think of the little user pamphlets that come with things like earbuds and can openers…), I’d be doing some strategizing.

, followed by a deliberate decision to go the other way (emphasis mine):

Quote

I’d ask myself: if the trends in MT, including neural networking, continue, am I going to be happy translating and post-editing 6,000 or 10,000 words a day for a couple of cents a word? Or would I rather look elsewhere?

In my own work, I’ve made a deliberate decision to work with a different kind of client. As an example, I recently started translating blog posts for a European environmental foundation. When I did the sample translation for them, they said, “Don’t treat it as a legal translation; rewrite it the way an American blog post should sound.” That’s the kind of client I’m looking for, because that’s a skill that MT doesn’t have.

 

Once again, I suggest that you read the whole article if you have time, but the main takeaway I took from it is that the translation profession is actually making a bifurcation at the moment. Some translators choose to become “translation software specialists” as Corinne puts it, and aim at somewhere like $20/hour,  perhaps for post-editing jobs —

Quote

I’ve recently talked to various beginning translators who’ve told me something along the lines of, “I currently work as a barista/nanny/museum guard/bilingual administrative assistant, making $9.25 an hour plus tips with no benefits or paid time off. So if I could make $20 an hour and work from home on my own schedule, I’d be thrilled.”

— while others will invest time and effort into self-growth and, ultimately, much higher rates and much well-educated clients — where “education” stands for “understanding that good translation is essential for their globalization, and that they cannot afford cheap solutions”.

 

I don’t see any conflict in this. I think this happened to so many industries, and it did not hurt the “top-end” vendors. On the contrary, it allowed them to clearly stand out among the general ordinariness and present themselves as an “elite” and hence raise their rates even more.

I also hold no grudge against those who choose to focus exclusively on post-editing, for lower rates and lower revenues. Actually I would argue that a really good post-editor can earn as much as a good translator. But, then again, this would be a totally different job — and a totally different story.

 

What do you guys think?

@Otávio Banffy @Virginia Monti @Paz Sepúlveda @Becky Pearse Romera @Patient Xavier Nong @Wordslinger @Faustina @Carinne Sá @Roxana Rivera and everyone :) 

  • Thumbs up 7

Share this post


Link to post
Share on other sites

I completely agree with Corinne. Things are going to change and we will need to adapt to it. The ones able to adapt well to those changes will thrive. It is just natural that things will change in several industries. 

  • Thumbs up 2

Share this post


Link to post
Share on other sites

In my humble opinion, @Vova, I think there is nothing like low rates and TMs swallowing us up.

The reason for my stand is quite simple: today, well educated clients (most of them being companies) know how important it is to hire the best linguists in order to translate their business into success in as many international languages as possible. And talking about that very aim of translation (without neglecting translation for domestic purposes), it is obvious that experienced linguists must be able to convey both message and emotions. But, as we all know, machine translation is not yet up to doing this. Even Neural Machine Translation seems to have a long way to go before being able to threaten linguists who are actually good at what they do, and who do it with so much passion and professionalism that they update their skills and knowledge on a daily basis, which is a good pace to challenge and even defeat our increasingly sophisticated machine competitors in the long run!

  • Thumbs up 1

Share this post


Link to post
Share on other sites

I've read the article, and I enjoyed it quite a bit. I'll be treating low rates and "robots" separately.

"Robots":

Firstly, machine translation isn't made by robots. That's semantics, but robots have a rather tight definition, and simply put, every robot is a machine, but a machine is not necessarily a robot. I could give the exact definition but that's unnecessary, I just wanted to bring this to attention.

I believe that machine translation might one day replace human translations, not just in generic translations but in all kinds, including poetic, literary, all sorts. I don't think this is coming any time soon though. It could be a few years from now, or it could be centuries. I also believe we don't need to worry about losing our jobs even in the event that that ever happens, for various reason. Some of them being that such advanced translation system would be very restricted and expensive.

Then, even if a machine is designed (or learns by itself) how to translate general materials with perfection, there will still be creative sense and cultural connections to be made in the translations, transcreation itself, and field-specific translations. Meaning, it would be even more complicated for a machine to be able to cover all those aspects, adding years and years to its design or learning.

Machines aren't coming to replace us anytime soon, if ever, in my perspective.

Though I should remember us all that we are already using MT for improving our translations. We can translate a lot more if we use MT systems, that's why we have these options in smartCAT. So, they may not replace us, and they may in fact create a better environment for us to work with. You could make a living by working as a Machine Translations Editor. I'm certain you could work with huge volumes using something like Lilt, or other similar systems.

Low rates:

I think Corinne made some pretty good points in her article about that. Just like I mentioned in another topic, the more niche your work is, the better (quality-wise) clients you should get, and the higher pay will come with it. I spoke about this in this topic:

Scraping for low rate jobs makes for a stressful and chaotic life, I believe. We should all find our niche at some point, or at least I think we should.

  • Thumbs up 2

Share this post


Link to post
Share on other sites

I'd say, aside from what has been said before, that machine translation is no different from the new tech being developed for medical purposes. They help a lot and can make a surgery that would have otherwise taken 7 hours, a matter of just a couple of hours, but the surgeon is still necessary and their expertise is still invaluable for the patient. We can and should use MT and TMs and all kinds of other tools available if this makes our job slightly easier, but that doesn't mean the need for good translators will fade.

Again, here we fall into the area of good and meh translators, because in the case of the latter there may be more evident effects in terms of job availability, but that's why we should always strive to be the best version of ourselves and offer things the machines cannot do, like cultural context, localization, emotion and humor, just to name a few.

  • Thumbs up 4

Share this post


Link to post
Share on other sites
1 hour ago, Otávio Banffy said:

I believe that machine translation might one day replace human translations, not just in generic translations but in all kinds, including poetic, literary, all sorts.

I disagree with Otávio on this one. Since the dawn of time, human beings have developed tools and technologies to make things easier and, from my perspective, MT is just one more example of this human effort.

As Paz says, they help a lot, but human beings are still necessary. So I don't see this as a threat at all. On the contrary, if MT means more job opportunities for translators and reviewers, then welcome!

Of all professions out there, I think we're lucky to have such a wide scope of possible applications of our knowledge of languages, aren´t we?

  • Thumbs up 3

Share this post


Link to post
Share on other sites

This is quite a unique situation where I agree with both points of view. Yes, our lives are changing with an advance of MT, ML, M-this-and-that, and machines are already here to do our work. And yes, machines will never be able to replace human translators completely. Let me try to explain why.

There are different planes of communication:

  • Machine to machine (M2M). This one does not even require a human language, and they've been talking to each other for the last 50 years.
  • Machine to human and human to machine (MXH). We use computers for... computing things, and talk to them daily through the Human Interface. However, there is not much they can really tell us (for now, at least), and there are things they cannot even explain to us.
  • Human to human (H2H). Now, here is the crux: most of the information we get THROUGH computers and machines comes FROM humans, either directly or via programming. If some human doesn't program the machine to act in a certain manner, it will just sit there and blink its pretty lights ("A machine that goes, 'Ping!'"). Translators are needed to translate things between humans, to be in this plane.

And now comes an interesting part. I am ready to admit it is possible for a machine to become as creative, unique, unpredictable, and awesome as a human being. What is going to happen then? We'll just get another human being with a need to communicate with us, humans. This also means it will be as limited as we are. Sure, a machine can learn all human languages, it's not very difficult. But what about culture? Arts, customs, religions, fallacies, fantasies, conspiracy theories are created by humans continuously, so even the most sophisticated machine will have to be helped (dab if you agree). Which means more work for us, translators. B|

Share this post


Link to post
Share on other sites

Personally, when it comes to specialization (as a 'weapon' against low rates), that's why I'm learning Japanese now. In Russian segment here, you have ~350 pages of translators from English to Russian, and less than 5 pages (last time I checked) of translators from Japanese to Russian. Now that's a good way to get 7000% boost in the probability to be picked by a client

Of course, I get it that there's far less work in Japanese - Russian segment, but still. Since I would still learn it (I love Japanese culture), it's a nice addition to my skillset

As for MT, I would say that, again, it depends on the language. Linguistically speaking, various languages just have different ways of working in their core. Most importantly, there are analytic languages (like English), and synthetic ones (like Russian), and there's a prevalent word order in almost any language (even in Russian, even though we don't like to admit it). Futhermore, there are unique traits that you need to consider when building an MT engine for that language (like an abundant amount of particles in Japanese). Long story short, you will have to adjust even a superb TM core to any specific language, and it would take quite a bit of time. So, if there ever will be an adequate MT to replace human translators, this will begin for English first, I think. Just because it's one of the most spoken languages.

So my Japanese knowledge will keep me in the biz for a little longer even then :)

  • Thumbs up 1

Share this post


Link to post
Share on other sites

I've read the whole article as per Vova's suggestion and going through it I said "oh, well things have changed a little bit since I get my Certification in 1998". 

Then I kept on reading and "voilà" I found this part: 

Quote

When I started in the industry (insert cranky old-timer voice) fifteen years ago, things were different. Even the mega-agencies generally paid non-peanuts rates, and most freelancers simply would not work for minimum wage. But lots of things have changed since then.

Yes lots of things have changed since then @Corinne McKay. In order to do a translation I had to go to the library to get as much information as I could then go home open 5 or 6 dictionaries I used to buy in Bologna, where I could find the best international book stores (1 hour by train from Florence). Now I sit down and I have to open 4 or 5 internet pages, drink a good espresso, open smarCAT and go! Not so easy, but much easier! 

In my opinion, MT can't replace heart, sweat, sleepless nights and sometimes tears that a translator put in every word she/he writes down giving a soul to the translation. We have a lot of good tools that make the life of a translator easier and, yes, I admit that tranlating is now maybe quite faster, but the quality is the same as 20 years ago if the translation is made by a translator who put passion, accuracy and professionalism in her/his work. 

This is my personal opinion.

  • Thumbs up 1

Share this post


Link to post
Share on other sites
3 hours ago, Faustina said:

In order to do a translation I had to go to the library to get as much information as I could then go home open 5 or 6 dictionaries I used to buy in Bologna, where I could find the best international book stores (1 hour by train from Florence). Now I sit down and I have to open 4 or 5 internet pages, drink a good espresso, open smarCAT and go!

Love this quote! I guess we often tend to forget that whatever the per-word rate is, what really matters (money-wise) is how much you earn per hour. And we used to spend many more hours for the same number of words back in the day :)

(My own cranky-old-voice story features me typing down something (I could not even loosely understand) in Polish, to my Polish uncle's uttering it letter-by-letter while translating it from Russian from some awfully Xerox'd user manual.)

Share this post


Link to post
Share on other sites

I have to say, I find @Corinne McKay's article reassuring. I feel that even though things are changing in the translation industry, I just can't see machines taking over our jobs any time soon. Along the lines of what @Paz Sepúlveda has said, if anything, technology will help us save time and perhaps force us to specialise more and become more technologically adept.

I highly recommend anything by @Corinne McKay btw! She does a great podcast with Eve Bodeux.

  • Thumbs up 1

Share this post


Link to post
Share on other sites

I may go a bit out of scope here, but I want to elaborate on that.

When I was in college studying robotics, labor replaced by robots was a discussion that frequently arose. There were good arguments to both sides of the spectrum, but a few numbers showed the truth. On one hand, we had the people who believe that no, robots are not taking away people's jobs because when a robot replaces someone, that someone is transferred to another sector where their skills are of use, and if they are fired, someone will have to be hired to take care of the robots as well. On the other hand, an industrial "peon" who gets replaced don't always possess the luxury of being able to find another job, and seldomly goes through the transformation into, say, a successful painter, and that's worsened when one robot replaces 200 people. You won't get 200 robot-fixers, and they won't be the same workers who were fired, in any way.

Sadly, that's the cost of technological advancement. The improvements in productivity, quality, production speed are unparalleled. But people get left behind in those changes. Their only alternative is to improve themselves into being useful for the corporations, or to create a business of their own. You can't shame companies for doing what's economically best for their already dwindling businesses, and you can't shelter all the workers left behind. Robots replace people in physical labor.

But machines haven't been restricted to physical labor.

Why do you need 7 makeup artists when you have Photoshop? Why do you need 60 scribes when you have OCR? Why do you need 10 accountants when you can have a single Excel boy in the office? Why do you need code analysts when you have Dreamweaver? ... Why do you need to hire 30 translators when you have 3 who use Lilt powered smartCAT already?

These tools aren't replacing the need for someone to operate them, but they are already taking away the need for too many people, or highly specialized people at that.

Then I come to the subject of intelligence. Our main argument why machines won't ever replace creative labor is that they can't feel or think, they don't have social experience or cultural understanding. They are not human. They don't have emotions.

But machines are not so different from us. We are not so special as we like to think. Our bodies may possess incredible technology, but machines are no less incredible.

Intelligence is defined by the capability to associate data, data which comes from the senses.

We have dozens, if not hundreds of senses. Sure, we have vision, hearing, taste, touch, smell, but we also have emotions, pain, a bazillion chemical reactions within our bodies which tells us all sorts of information. Well, machines have all that, and more. A machine doesn't need to be restricted to a single body, it can have infinite eyes. In order to taste, you only need to give it a tongue, and that tongue can pick up interpretations our organic tongue cannot. In order to hear, you simply need to give it a microphone. In order to smell, its tongue does the trick. It can pick up touch (touch-pads and pressure plates), and it can also feel electric fields.

While you can't typically tell your heartbeat without focusing on that, a machine can tell its core temperature, fan rotation, memory usage, power consumption, the keys being pressed as input, and it doesn't even need to blink while it's at it.

So you're asking, "if machines are so incredible, why they don't compare to us?" That's because we are running on a very efficient reaction-based organic system, and they are running on synthetic materials designed by us, not by nature's reaction-based system. We are intelligent because we have a system that creates thinking through the reactions in our bodies.

Plants are intelligent, hell, amoebas are intelligent, and they don't even need a brain for that. Machines are not intelligent because we haven't given them the capacity to think, yet. They can make incredible calculations, as far as we teach them how to do it, but they can't learn for themselves, because we haven't taught them how to learn. But machines are way more powerful than we are in regards to the control they can have of themselves.

Once you manage to create a system, a body, which can learn through its senses (and it doesn't even need to be an incredible body at that), you'll have an intelligent being, but different from the designs of nature, you don't need to "restrict" it with pain, need for sleep, hunger, or else because its sensations can be controlled and interpreted by whatever standard you give it (or it gives itself), and from the point of being alive it has all the potential in the world to learn everything we know, including language, culture, the understanding and capability of feeling emotions, and everything we humans have, and it won't be dying of old age any time soon, nor it needs to necessarily have an ego.

We have a tendency to put ourselves on a pedestal, think ourselves incredible due to all humanity has achieved and what not, but truly, we are incredible, we are just not so special as we like to think. There's a reason for us to be the way we are, and once you topple that, you have an understanding of what we are, and once we can define ourselves, we look no more than a very efficient machine.

What does all that has to do with translation? -- Everything.

When you have a machine that is alive and can learn, it is unrestricted to learn everything we think creatively and dominate it in a scale we can't hope to reach ourselves, because we can't design ourselves forward, and it can.

When that happens, if it ever happens, we will have reached the robots of today, replacing physical labor, but this time replacing all kinds of labor.

I won't pretend to know what is it that we have in store for us, but I do want to explain that we are not so special as we think we are, and that being replaced is quite a possibility.

Now, be at ease. With all this said, I believe you have no reason to worry, because such a thing is probably too far away from ever happening that you will likely not even hear of its prelude. We could even be done as a species before that ever happens. I just wanted to explain how that is possible.

 

  • Thumbs up 2

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×