AI or human? Why the future of interpreting is hybrid
A few weeks ago, KUDO launched a new integration with Google Meet, allowing businesses to add AI-powered speech translation and captions in more than 80 languages to meetings on the platform.
Meanwhile, Interprefy announced Speaker Labels, a new feature for multilingual captions which identifies individual speakers by name.
It’s clear that AI translation is increasingly accessible for companies, but human interpreters aren’t redundant just yet. I think we’ll increasingly see a hybrid model, where organisations choose the level of language support according to the complexity and risk of the meeting.
There are situations where AI makes perfect sense, such as a relatively straightforward company presentation with one person addressing an audience.
For example, it can be a useful and cost-effective solution for company town halls or straightforward one-way presentations. For EWC meetings and social dialogue, however, I would be much more cautious: the nuance, interaction and potential consequences of misunderstanding mean that human interpreters remain the right choice.
Perhaps the organisation would like to make the information available in several languages, but simply doesn’t have the budget for teams of human interpreters.
In that situation, AI translation can provide access that might otherwise not exist at all.
It may not be perfect, but sometimes a less-than-perfect translation that allows somebody to understand the essence of a presentation is better than no translation at all.
However, AI still struggles with technical terminology, local idioms, cultural references and other linguistic nuances such as sarcasm or humour.
It’s also not ideal for complex discussions, where several people begin asking questions, interrupting one another and moving rapidly backwards and forwards between different topics and languages.
Then there’s the question of risk. If you’re discussing a routine internal update, an imperfect translation may be acceptable. But the consequences of a misunderstanding could be enormous for high-stakes, highly technical discussions like contract negotiations or restructuring consultations. Those are the situations where professional human interpreters remain essential.
From an agency perspective, even when companies are using AI translation for meetings, they will still need support to plan and run the meeting. Recently I’ve been advising businesses on the right platform for their needs, and ensuring they have the support in place to ensure the technology runs smoothly and troubleshoot any issues which crop up.
Our role as language professionals is no longer to insist that human interpreting is always best, but to help clients understand which type of interpreting – human or AI – is right for the situation.