The final instalment in AIIC UK & Ireland’s “Artificial Intelligence and the Interpreter” webinar series took place on Friday 15 January in the form of a panel discussion. The aim was to draw conclusions about the extent to which our profession is under pressure, to identify how AI solutions will impact on our profession in the coming years, and to consider how AI is perceived by both interpreters and users of interpretation services.
 
We were joined by Thomas Jayes, Head of the Strategy and Innovation Unit at DG LINC in the European Parliament; Antonio Paoletti, Head of the Meeting Services and Interpretation Section at the International Maritime Organisation (a UN agency based in London); Naomi Bowman, the CEO of DS-Interpretation, Inc.; and the two automated speech translation experts from our first webinar, Dr Jan Nieuhes from the University of Maastricht and Dr William Lewis from the University of Washington, who were on hand to provide advice on technical questions.
 
The chair of AIIC’s UK & Ireland Region, Monika Kokoszycka, set the scene by summarising what we had learnt in the webinar series so far. She explained that the objective of the webinar series was to arm ourselves with knowledge about AI, so that we are informed about what AI can offer, and can advise our customers accordingly. The advent of AI-powered speech translation solutions is a twist in the history of our profession and will change the market, but it does not have to be a fatal blow.

The panellists then explained how their institutions or private-market clients were already using AI. Thomas Jayes reported that the European Parliament was developing a number of AI-powered projects, including an initiative to make parliamentary debates more accessible (speech-to-text solutions), and is considering further use of AI to support the interpreters (for example, by providing information about speakers), and help with recruitment and programming of interpreters in their complex organisation. Antonio Paoletti agreed that there was no question of machines taking over from human interpreters, identifying speech recognition as being the main obstacle, particularly in the UN, where 60% to 70% of speakers are not using their mother tongue. This is compounded by the fact that humans do not naturally speak in full sentences or perfectly accurately. The panellists agreed that speakers would not be prepared to change the way they speak in order to be interpreted by a machine, nor would they be capable of doing so: speakers might be experts in their specific fields but not necessarily expert communicators.
 
Naomi Bowman pointed out that customers’ perceptions are often very different from reality: clients assume we are using AI and have done for years. It is therefore vital to educate clients about the actual capabilities of AI in interpreting. Will Lewis confirmed Thomas Jayes’ comment that at present AI is of greatest use for increasing accessibility by adding automatic translation to already existing live captioning services. Mistakes are made, agreed Dr Jan Nieuhes, the cost of which is too great. He believed that AI would be used primarily in new markets and not in settings where human interpreters currently work. Naomi Bowman gave an example of a conference where an AI-powered speech translation solution had been used successfully, partly because the event was very tightly controlled and the speakers had been trained to deliver their comments in a particular style. The outcome was less good than human interpreters (a longer delay, errors, less comfortable to listen to), but had still exceeded the customers’ expectations. As their alternative was to have no interpretation and the conditions were conducive to AI success, this was considered an appropriate use of AI to provide language access.
 
The panellists were unanimous in believing that AI is not the greatest threat to the interpreting profession. Thomas Jayes believed that the greatest threat to conference interpreters continues to be “no interpretation”, because interpretation is too expensive, there are not enough interpreters available or because of the predominance of Globish. He pointed out that remote interpreting had previously been seen as a huge threat to conference interpreters, but is now perceived as a lifeline. Naomi Bowman believed that the greatest threat to the profession is the commoditisation of conference interpreters due to a lack of differentiation between conference interpretation and other types of interpretation.
 
Here, client education was perceived to be key, and webinar attendees were invited to contribute to a list of advantages that human interpreters have to offer over automatic speech translation solutions. The results included that human interpreters are better at taking account of non-standard pronunciation and disfluency, can get on the speaker’s wavelength and look beyond their words to their message and argument, have an awareness of culture and register and empathise with the speaker’s emotions and state of mind, thus adapting their tone accordingly. The full crowdsourced list is available here.
 
How can we conference interpreters best respond to the challenges facing our profession? Thomas, Antonio and Naomi made three suggestions: we need to be adaptable and flexible; we need to embrace new technologies rather than shy away from them; and we need to be more aware of the context in which interpretation takes place, particularly now that the market is in flux.
 
In conclusion, the speakers agreed that AI has its place in interpreting, particularly for improving accessibility. While “AI is about breadth, human interpreters provide depth”, as Will Lewis put it. The language industry is growing - indeed it even grew during the pandemic - and AI is contributing to this growth by making people more aware of linguistic diversity.
 
Our final poll of the series confirmed this fairly optimistic picture: 75% of participants believed that human interpreting has a future.
 
The AIIC UK & Ireland organising team - Françoise Comte, Louise Jarvis, Monika Kokoszycka, Stefanie MacDonald, Deborah Muylle and Monica Robiglio – would like to thank all the speakers who have generously contributed their time and expertise to this webinar series. It has indeed been an odyssey for all of us, but we emerge better informed about the reality of the impact we can expect AI to have on the interpreting profession and therefore better able to hold our own in this brave new world.

Some footage from earlier webinars in the series can be found on AIIC's YouTube channel.