Panel discussion: AI in the language industry: threat or opportunity?

5 panelists joined us from Israel, Canada, the Uk and Holland to discuss the present and the future of the linguistics space in light of machine translation, voice-to-speech technology, language learning apps and AI advancements.

Martina Abagnale is an English and Dutch into Italian translator based in Amsterdam, the Netherlands.

Mariam Dar is a computational linguist and forensic analyst. She has been working for industry-leading companies in AI and Machine Learning.

Mark Ghannam is a Solutions Engineer at Labelbox. He helps data science teams and engineers to build scalable AI solutions faster in all industries. 

Yael Sela is a Hebrew Language Manager and MT Quality lead for Google’s Localization team – she works on their Machine Translation (MT) Implementation project.

The panel discussion was moderated by Richard Howeson, Founder and CEO of uTalk, a London-based language learning company.

Mark pointed out that the more specialised you are, the more difficult it is for AI to do what you do.

His analogy on whether we as language professionals should fear losing our jobs was thought-provoking:

If AI helps a doctor to analyse all the X Rays, that doesn’t mean the doctor will be out of job. It means the doctor will be able to see more patients.

Therefore, the question we should be asking is this: AI is automating this part of my job. So how can I offer something on top and diversify what I offer or expand my outreach?

Mariam mentioned that whatever is unknown looks scary. When something is new and evolving, it calls for learning new skills. And that means having to stretch our comfort zone as linguists.

Or as Yael put it:

A kick in your behind pushes you forward.

Hebrew saying

Martina embraces machine translation and uses it as part of her work on a regular basis. She pointed out that translators are not scared of AI per se. They’re scared of companies who don’t want to pay them anymore and of getting less work.

Richard then shared his experience of working with 2 translators, an editor and a voice actor when creating their language corpus for uTalk’s app to ensure quality translation. And he says that the human element is always essential in this process to make sure everything is accurate and culturally appropriate.

Yael raised the point that the translation and localisation industry is worth almost 60 billion USD a year and it’s growing rapidly due to global expansion of companies.

And everyone agreed that manual, repetitive tasks will be replaced by AI. 

AI is an opportunity to ditch the repetitive work of translation and shine where people have added value. And where, by the way, we can get paid more. – Yael

An example for this from Richard: he thinks that language teachers in the future will be more like language learning managers. As he puts it, their uTalk app would never make someone fluent. They’d have a lot of fun learning some basics which will help in having more meaningful conversations with the teacher. So language teachers will have the pleasure to teach the advanced side of language and not necessarily the basics. 

Martina points out that copy can now be automatically generated by tools like copy.ai. But that soon, humans will be able to distinguish between human-generated content and machine-generated one.

Mariam then spoke to how voice technology is changing the landscape and is creating many new jobs. But due to the ambiguities in the language, and because machines can’t read between the lines, humans will always be needed.

So, what will the future look like?

Mariam sees loads of opportunities emerging for health practitioners  including speech therapists. Diagnostic tools can now help to diagnose speech and language delay problems early on. 

Martina thinks people will be better problem solvers. As she said, quoting Mark: AI works on a garbage in – garbage out basis. But humans can intervene and raise awareness of any garbage going in, in order to avoid communicational or cultural disasters.

Yael mentioned that raw machine translations will be supplemented by post-editors who will ensure that the translated work is accurate and culturally sensitive too.

And people like us, language professionals will be constantly needed to train the machines and analyse data in the first place.

So, we all agreed that the future is very much human.

About the Panelists