Outed by AI? How to Right the Wrong

“Your Honor, why don’t we have the interpreter read the script generated by Zoom?” This was a question that came up in one of my latest remote hearing cases. If that was not enough to surprise many of us who are court interpreters, the judge’s answer might have done so: “Sure, let me ask her… madam interpreter, would you prefer reading the script generated by Zoom?”

In addition to surprise, I was taken aback, dumbfounded, and a little angry. In my experience, up until then, such a question had never arisen in a legal setting.  As you can all imagine, I politely declined the offer, and I put my job description and ethical responsibility as a court interpreter on the record. I was prepared to be standing up to a challenge rather than doing injustice to the LEP by taking the easy road and reading the computer-generated script.

We cannot deny that we all benefit from the ingenuity and ease offered to us in the form of AI—self-driving vehicles, content creation, or even assistance with medical diagnosis. But let us not forget that AI is created by humans to assist, not to replace. However, a question such as this judge’s is the reason many of us are terrified that AI will soon be doing our work at a much cheaper rate.

With the promising age of AI, the world is set for a shift in how much can be done daily, with more ease and at a reduced cost. It seems that the interpreting field is no exception. Discussions are posted daily about how AI can improve our productivity. Yes, I get it. AI is wonderful. But is it really? And more pertinently, will I be replaced?

A recent post on LinkedIn by Julia Poger was a good reminder to us all that we should NOT be in competition with AI. In Poger’s words, AI is a cheap magic wand that (potential) clients are using as a solution to cut costs. The courtroom incident above may not fall into this category, but nonetheless, it was an idea that was entertained (I tend to think that the attorney was using the interpreter as collateral damage, but that is a topic for another day).

Personally, I see AI for now as a mediocre interpreter who can certainly provide word-for-word interpretation for clients. But what about context? AI can certainly do a better job than a bad interpreter. But is that good enough for certain clients? Some may say that it’s cheaper, but we all know that cheaper does not always mean better. You get what you pay for these days, and in my experience the interpreting field is no exception. Good interpreters invest a great deal into their career—this includes time and resources to acquire and maintain or improve their skills, or to travel… and often there is significant spillover into their personal lives.

Many of us are no strangers to Google Translate, including myself. I have used Google Translate to help me while travelling in a country where I cannot speak the language. I have used it out of curiosity, to compare it to my own translation. These experiences have shown me that AI often lacks context. It’s robotic. It’s mechanical. It’s rigid. What AI cannot offer, at least not yet, is the nuances that a human can have in interpreting.

For example, in early October, I was working at an insurance conference in Manhattan. The agency offered to pay me an additional $20 to put together a glossary of insurance terms for the event. It was an offer that both my booth partner and I politely declined. On the day of the event, a glossary was placed in the booth for us. It was quickly obvious to us that the glossary was machine translated. Don’t get me wrong: we must give credit where credit is due. AI did a wonderful job on technical terms like agent, premiums, insurance policy, etc.… But it failed miserably when it came to terms that needed to be in a context to get the appropriate meaning.

For example, “elite” was translated as ผู้ลากมากดี, which is equivalent to the term “aristocrat.” While it is the correct synonym in the English language, for this specific context, the word “elite” (in its English acceptance) is normally used to describe an elite group of customers. “Dynamic” was translated as เคลื่อนที่, which is equivalent to “moving around.” Again, an acceptable term, but not for that context.

Lastly, “adopting” was translated as รับบุตรบุญธรรม, which means “to adopt a child.” Need I say more?

Or consider a more down-to-earth instance: I was in a meeting with a blind individual who described the wonderful world of using assisted technology in mobile-application format on her smartphone to help her read a text out loud.

“It’s useful,” she said, “especially when I check in to hotel rooms. I get to know which bottle is shampoo and which one is conditioner. But that’s because I am aware that those products are placed in the bathroom. Otherwise, you cannot trust them!”

She went on to explain to me how once she was checked into an Air BnB she had never been to before; an unfamiliar environment.

“I found a tube in the owner’s junk drawer, and the darned application told me that it was face wash.” Fortunately, she used a secondary method/device to verify the reading to find out that it was, in fact, wood polish! Another AI blunder showing that it has a ways to go. Perhaps the new world is not here quite yet.

Poger’s post also reminded me about the changes that came about when Covid hit in 2020. The world went into lockdown, and Zoom became the hero of the day. Many of us had to re-strategize our marketing and adopt different service methods. But it worked, and here we are in 2023, offering both in-person and remote services. So, what makes this time around any different? Another challenge, another bump in the road.

It’s that time again, my friends. Rather than panic before the threat, we must adapt to this fast-changing environment that can make one feel like a character in a 90s sci-fi movie. While the choice lies in our clients’ hands, we can educate the client. We can switch our mindset and move away from the fear of competition to better adapt to the impending invasion of AI. Think about the ways that we can use AI as a tool to show our clients the value that we bring to the table. What we can offer as interpreters are human judgement, expertise, and personal relationships that are not possible with AI. Keep working hard and know your worth! Nosce quanti aestimanda sis.

Ann (Jiraporn) Heath-Huynh grew up in a bilingual Thai-English household, using both languages in day-to-day life. Having lived on four continents, she now calls the U.S. home. Following the birth of her daughter in 2010, the chance to work in the language field afforded her an opportunity to change career directions; what began as a part-time job became a career that she is passionate about. After being added to the Maryland Judiciary’s roster of interpreters in 2015, many opportunities opened up to her, eventually leading to Department of State Conference Interpretation for Thai and English. Although she works mostly as a conference interpreter, she has always considered Maryland courts to be her home. Whenever an assignment is offered, she is always pleased to accept it and proudly wear her first-ever interpreter’s badge as a Maryland Judiciary Court Interpreter. Contact: ann.h.huynh@gmail.com

Featured photo: from “Así funciona el piloto automático de los aviones (y por qué ha tardado en llegar a los coches)” by author Ibáñez at Hablemos un poco de todo, under a CC BY-NC-SA 3.0 ES license. Text-body photos: from “Las posibilidades de Arduino en el mundo profesional” by Ruben Beiroa at VERMISLAB, under a CC BY-SA 4.0 license, and again from “Así funciona el piloto automático de los aviones (y por qué ha tardado en llegar a los coches)” by author Ibáñez at Hablemos un poco de todo, under a CC BY-NC-SA 3.0 ES license.

11 thoughts on “Outed by AI? How to Right the Wrong”

  1. Barbara W Considine says:

    Wonderful, informative piece, Ann. Thank you!

  2. Thank you so much for sharing your experiences, Ann 🙂 Your anecdotes are fascinating (and rather humorous). This really shows the relevance and limitations of AI.

  3. Excellent points Ann. Thank you for bringing this alternate point of view on AI.

  4. Nurit Shoham says:

    Loved your article and how you intelligently described AI’s deficient intelligence. I’ve been experimenting with AI for various uses in language instruction (Hebrew) and so far am very disappointed and often amused.

  5. David Proano Celi says:

    Thank you for sharing your experiences and for your perspective. While AI does have limitations seen throughout all related topics and professions in the AI landscape, my experience has been quite different as far as quality of translation, in fact, there are different companies who offer different AI products (ChatGPT, Bard, etc.) where I found that there are exceptional levels of translation found with different machines, hence, results may vary. Taking in consideration two points: One is that machines will have more data on languages such as Spanish for training than others (rare languages). But this too, will most likely improve with time and improved models with more data becoming available for machine training in the future. And two, interpreters (in general) are more often having interpreting related experiences where AI is somehow involved in one way or another at different levels in the interpreting profession. My guess is that this is just the beginning of what AI is able to do in relation to language interpreting, transcription, speech to text and so on, and I wouldn’t want to speculate or guess what the future will be with AI in relation to language interpreting.

    1. Observer Editor says:

      No one can predict the future; of course we could have a “worst-case” scenario in which, say, in two years, voice-to-text software makes a breakthrough and can make impeccable and instantaneous transcriptions, in spite of all kinds of audio and accent problems, and then from that instant transcription DeepL, ChatGPT, Bard, and other smart MT tools produce instant subtitle translations that are adequate in 99% of cases. But seeing as how *slow* progress is with human beings (I’d say the majority of meeting participants in conferences still have no idea how interpreting works) I find that very, very implausible. And even if it did happen, what does that mean? That we all starve to death? No. We just reinvent ourselves and find something else to do. Wouldn’t be easy, not by any stretch of the imagination, but everyone can do something useful and work around changes that seem to be purposely designed to make our lives difficult…

    2. Jiraporn Ann H Huynh says:

      Thanks David for your input. I can’t deny that I agree with you and just had a very fruitful conversation with my Russian colleague last night about this.
      We both agreed that AI has come a long way in translation but unfortunately, not yet with simultaneous interpretation. It has come a long way though and now it all comes down to what your client wants. In certain areas, like diplomatic and high level conferences, I don’t think that protocol would head into that direction for many obvious reasons. Web AI translation services like DeepL does a pretty scary (good) job at being accurate with the translation, again, it can be an amazing prep tool when time is short. I am not opposed to using AI at all, it’s a matter of preference I think but to see the trend invading the courtroom space frightens me. It is equivalent to me getting back to the attorney and tell him that I can use an AI application to assist me in a legal case instead of having a real trial attorney. For that specific situation, in my opinion, it can create legal loopholes that can lead to other problems. And to see how people want to use it as a quick fix makes me sad about the shifting values of society. I agree, the future of AI working in SI and CI level interpreting is coming, just not quite yet.
      I love to hear about the experience of others and start up a constructive exchange like what we are having now. Thank you for stopping by and spending the time to be part of the conversation!

      1. Observer Editor says:

        Exactly! Except I would argue the same even in translation.

        AI can arguably find medical, legal, or financial information far faster than a human ever could. But how many people entrust their medical issues solely or even mostly to AI? Only the foolish or the daredevils. What company or individual entrusts their accounting or their legal issues to ChatGPT alone? Come on. Artifical intelligence can in theory do all those things better than a human can, and yet people still – and rightly so – systematically turn to a human when it’s anything serious. There may be some wild minds who see this as a flaw and want to change human nature, but they’re on the wrong side of history.

        I’ve asked medical and legal questions of ChatGPT before, and it ALWAYS (ad nauseum) insists that you consult with a licensed or certified professional, beyond the (albeit impressive) few paragraphs it writes for you.

  6. Jesse says:

    For some reason this reminds me of a (somewhat dark-humored) joke where two buddies are walking through the woods and come across an enraged grizzly bear… one of them starts running as fast as possible in the opposite direction and the friend tells him “Don’t bother running, there’s no way humans can outrun a bear!” to which the sprinting friend retorted “I don’t need to outrun the bear, I just need to outrun you!” In a similar manner, sometimes we err thinking that AI needs to be absolutely perfect before it can replace any of us (speaking of any field; be it interpreting, driving, reading x-rays, etc) when in fact, it only has to be better than most of us before the allure of an instantly available, OSHA-free, money-agnostic, non-complaining, indefatigable worker makes employers gravitate to that new (artificial) kid in town. We should use the same outrunning-the-bear approach to overcome (or at least survive) this predicament… strive to be at the top of your craft so as to not be in the group that is so easily replaceable. I like to tell clients that balk at a quote “If you think hiring a professional is expensive, just wait until you hire an amateur.” The caveat with such a statement is that it rings hollow unless you back it up with performance that truly does bring value to the process. The AI bear is scary, I’m not going to sugar-coat it, my best advice to colleagues is to do whatever it takes to stay running at the front of the pack.

    1. Observer Editor says:

      That is true, and it would be true even without the towering AI grizzly bear because there always have been and always will be, in any profession, people who are better than you and people who are not as good as you are. You always have to be at the top of your game. AI can help with that! Why the heck not. If there’s a war, and the enemy army is about to bust out tanks while ours uses horses, the smart thing to do is, I dunno, steal a tank, figure out how it works, and build an army of them as fast as you can. Putting our heads in the sand won’t work, and so yes we should stay at the cutting edge.

      Another example from aviation (slightly modulated from your quote)… airliner maintenance is unimaginable expensive (and airline profits are always razor thin), but I saw a poster that said: “You think aircraft maintenance is expensive? Try having an accident.”

      So far I have direct experience of meeting participants turning to AI solutions, they quickly coming running back to the human instead. It was a high-pressure IT project in the hundreds of millions of dollars. I can imagine that political, military, judiciary, or medical matters are orders of magnitude higher in terms of pressure. Again, doesn’t mean AI won’t advance by leaps and bounds in the next few years, but okay, just pay attention to it, then. How will worrying about it help us?

    2. Jiraporn Ann H. Huynh says:

      My point exactly. Thank you for sharing your analogy and advice.

Leave a Reply

Your email address will not be published. Required fields are marked *