According to scientists and developers, artificial intelligence is not meant to replace conference interpreters.

Dear Colleagues,

As you begin to read this post, I assure you: This is not another article on artificial intelligence (AI). We all know there are plenty of those, and we are tiring of reading, and hearing, the same.

I attended many professional conferences lately, and I tired of listening to interpreters and translators presenting on artificial intelligence. Some conferences were even worse: They included, as part of their programs, presentations by the salesforce of AI companies. It did not take me long to question why these events, designed for us to learn about something we were not familiar with, were inviting people like us: interpreters and translators, who, granted, like technology and have acquired some knowledge on how to use the latest products, to speak about artificial intelligence, algorithms, chat GPT, etc., presented them as experts and let them run with their ideas about technology, ethics, and the law. I soon learned that I was not alone. Many of my colleagues were wondering the same. The confusion grew even more when some professional associations, and businesses gave AI companies’ marketing departments full access to these events and publications. Said businesses disseminated information about the possibilities AI was bringing to the market, and told stories of our demise as a profession. The propagation of these ideas was particularly harmful when fellow interpreters and translators turned soldiers of fortune and validated what the salesperson was saying. Platforms and agencies, dazzled by the potential of a future with higher profits and robotic interpreters, began a public relations campaign to convince the market, and the general public, that AI interpreting was as good as a human rendition… heck, it was better as computers do not get sick, take care of a child or parent, and do not demand work conditions such as a limit to the hours in the booth for example.

I was pleased to learn that the Private Sector of the International Association of Conference Interpreters (AIIC PRIMS) had organized a conference on artificial intelligence and conference interpreting where instead of getting together with our colleagues to talk about issues we know little about, we would hear from those who truly know the subject: the scientists and developers of the technology that gave us AI, those who organize the events where we interpret, and the lawyers who specialize on privacy, intellectual property, and artificial intelligence.

Much has been said of the immense quality of this event and its sound success. I invite you to watch the videos and read the reports you can easily find online. Instead of repeating what has been already said by some of the top interpreters in the world who attended this meeting in Bali, Indonesia, I will share what I took from the event, the conversations with my colleagues between sessions, and what I believe we should do as a profession. Among the many chronicles of the PRIMS meeting, and read and watch as many as you can, I suggest you read the summary by my colleague Veronica Perez Guarnieri on LinkedIn https://www.linkedin.com/posts/veronicaperezguarnieri_ai-interpreting-primsinbali-activity-7152119149870141441-sye7/?utm_source=share&utm_medium=member_ios and watch Seth Godin’s video on YouTube https://www.youtube.com/watch?v=AkkoA9j7BOU&t=4s

I left Bali with these lessons learned:

• Artificial intelligence and other technological accomplishments are here to stay. It is inevitable; it has happened throughout history, and instead of fighting it, we must embrace it and use it to our benefit. Seth Godin made it clear in the video (see above). Before record players there were many musicians as there was no other way to have music. When the phonograph came to be, people had the option to listen to a recorded piece by a world-class violinist, pay more money and have a virtuoso perform live, or hire the same mediocre violinist they were hiring before the phonograph. Many lost their jobs, they abandoned the profession, but the great ones, those who could play and sound different from the recording thrived, and still do despite digital recordings and the availability of music to be downloaded to our telephones. The same will happen to interpreters in general. We must work on two things: Improve our skill so we can apport those nuances nobody else can, and master the technology so we can use AI as a tool and be competitive.
• Artificial Intelligence will never replace conference interpreters. The people who know the subject and are concerned with science, research, and technology at the top companies worldwide told us their goal is not, and never was, to develop a product that could interpret a conference. They explained their mission was to develop a product that could do the functions of a public service interpreter to provide the service when no human interpreter was available, and this way avoid a tragedy or a miscarriage of justice. For example, AI can, and will improve, interpret a healthcare encounter when an individual breaks a leg, or suffers a heart attack in the mountains of Alaska, or small village where no human can assist the medical staff. The same is applicable for a court first appearance hearing in a remote area, or a holiday, and this is the only way that a person can bail out instead of staying in detention for days for a case that would let them post bail immediately if this happened in an urban area on a workday. The idea was not to get AI to interpret a jury trial. AI interpreting for conference was not even on their radar because linguistic and cultural complexities would render it useless.
• AI knows what it is taught. This has many consequences, but in particular two that directly apply to conference interpreting. It was explained that it is (relatively) easy to develop algorithms that let AI go from written word to speech (this is how preliminary court hearings and healthcare encounters based on written text can be interpreted by a machine); but it is extremely difficult to go from the spoken word to the AI rendition, thus it is impractical for conference interpreting. The second consequence concerns language combinations. There is a wealth of knowledge, and algorithms in English, and a pretty robust bank in other Western European languages such as Franch, Spanish, Portuguese, German, etc., but knowledge and algorithms are scarce in many languages of lesser diffusion in the West. A healthcare encounter that AI could interpret from English into Spanish could not be interpreted from Pashto into Farsi.  When referring to the written text the experts made a remarkable statement in my opinion: They claimed that unlike conference interpreting, translation is “dead” as a profession as the written text algorithms continue to develop in many languages. They are partly right. I can see the day when basic documents will be AI translated, but I think the same principles that apply to conference interpreting apply to literary translation. I am not a translator, but I would tell translation students to forge ahead and continue their professional formation as translators if their goals are to translate complex text. Nobody should go to college to translate birth certificates for the rest of their lives.
• Scientists are working on developing artificial intelligence to be a tool conference interpreters apply to their practice. This they take very seriously and the technology must have two clear goals: (1) Be relevant to the work of a conference interpreter, like the possibility of getting figures on the screen instantly converted from the English to the metric system, seeing acronyms, numbers and names of places or entities on the screen in real time; and (2) AI functions cannot increase the conference interpreters’ cognitive load. A new function that requires the interpreters’ attention and the manipulation of additional elements would not be developed.
• Developers and scientists have asked their marketing departments to avoid enhancing the functions and reach of AI interpreting. It is the salesforce, not the scientists, who have introduced to society the idea that artificial intelligence can replace humans in conference interpreting, and the salesforce has recruited interpreters to advocate for these functions. This sector of the AI world conveniently leaves out AI’s limitations, and fails to mention that AI is not to conference interpreting but a tool, just like the booth, the headset, and the computer. Those in the meetings, incentives, conference, and exhibitions (MICE) industry wholeheartedly support humans in the interpreting booth, they know what it takes to interpret a conference, but they also warn about the organizers of marginal events, those who are not mainstream, and out of lack of knowledge and desire for profit, schedule these fringe events after taking advice from the salesforce.
• Before dealing with a platform or an agency, conference interpreters must be protected by a contract that covers their intellectual property and confidentiality rights. A bullet proof agreement that guarantees their work product will not be used to develop algorithms, profit from the repeated use of their rendition, and releases them from all violations to the duty of confidentiality by the platform, organizer, client, or conference attendees. I also rescue the recommendation to those who inadvertently allowed these entities to use their work product in the past to cut their losses, and instead of devoting their time and money to litigate these past assignments, destine those resources to a good lawyer to prepare a good contract for future work.

It is our job and duty to protect ourselves, our profession, and the public by acting professionally, and explaining the clients that AI was not developed for conference interpreting, that it belongs in public service interpreting, and it is the people in marketing, not the scientists, who erroneously claim so. Please learn as much as you can, become a unique interpreter who cannot be replaced like the bad violinist, but always listen to the scientists, not to the salesforce or their surrogates.

This site uses Akismet to reduce spam. Learn how your comment data is processed.