Hello, and welcome to my first blog post in over a year. I don't know if people still read blog posts in 2023, but here we are. I'd like to take the opportunity to share some of my thoughts on a ubiquitous subject: artificial intelligence in translation. I will try to keep this post short and sweet.
First off, I'd like to state that I have yet to hop on the Chat-GPT train. Call me paranoid, old-fashioned, a Luddite, what have you, but I don't see how it's really "free" when you go to the website and they ask you for personal data, including your email address and phone number, to use their service. Now, I know they probably have a bajillion security protocols in place, but they want that information for a reason. They are not just going to let millions of people use their service without something in return.
That is not to say I won't start using Chat-GPT or some other AI platform in the future. However, I have yet to see a good enough reason to adopt it into my work routine. Sure, I have seen some impressive tutorials, say, on how to use Chat-GPT to compose cover letters from résumés to save time. I just don't see the need. I am creative enough to come up with my own cover letter highlighting my strengths. By the time I'd have finished editing the AI-generated content, I may as well have started from scratch.
Maybe eventually. It's hard to give an estimate on when this will happen (if it will happen). I am honestly not too concerned, even if it will inevitably squeeze the market in the long term. Companies who have the resources (and foresight) will come to realize that, although human translators are slower and more expensive, what matters at the end of the day is the quality of the translation. In many sectors, the quality provided by machine translation will be "good enough," such as in technical or legal translation, or anything else that tends to be highly predictable and/or repetitive. And that's okay (considering your main work isn't in one of those fields). However, in sectors that require more creativity, such as marketing or tourism, machine translation will be too stilted and/or literal for years to come to be effective.
Industry leaders have expressed that interpreting will be more future-proof than translation. I agree with this assessment. As a subtitle translator, I have come across time after time instances where machine translation fails to accurately translate idioms or regionalisms. Read my post "Why Not to Use Machine Translation to Translate Subtitles" for more info on that.
People tend to speak differently than they write, especially in informal situations. For example, in many Mexican reality shows I work on, the Spanish word "pedo" gets thrown around in a variety of different situations for a variety of different meanings. It typically means either "drunk" or "problem," but has taken on a myriad of other meanings as well. In most other Spanish-speaking countries, "pedo" means "fart." As such, machine translation tends to translate it as such. You could see how this could be problematic.
This is the trickiest subject. There are several startups out there hellbent on putting voice actors whose primary income comes from dubbing out of work. They are doing this by "cloning" content creators' or podcast authors' voices and then using AI to translate their voices into a different language while (allegedly) maintaining the tone, timbre, intonations, etc. of the original voice.
I find this use of artificial intelligence unsettling and worrisome, but I guess I shouldn't be surprised. I am glad I don't offer voice-over as a service, but my heart goes out to those who do and may soon be put out of work due to this opportunistic "advancement." Of course, this raises all sorts of legal questions as to copyright and likeness (not unlike the issues at the heart of the ongoing actors' guild strike). But at the end of the day, I am just as skeptical as I am pessimistic about all of this.
Artificial Intelligence in translation is here to stay. While it may not be necessary to embrace it, we will, at the very least, have to acknowledge it. Translators will have to get more creative in differentiating themselves from not only their peers but machine translation options as well, which may be more attractive at first glance because of their lower price point and faster turn-around times. We have already seen the advent of the infamous post-editing of machine translation, a sort of "best of both worlds." I think this may be a viable "win-win" situation for both clients and translators, but only in fields where it makes sense to do so and only if the translators/post-editors are fairly compensated. Although I honestly do not have an answer as to an ideal payment model that would accomplish this.
What are your thoughts on artificial intelligence in translation? Let me know in the comments!