Automatic speech translation (AST)
It’s true that so-called artificial intelligence (AI) – in the guise of tools like ChatGPT, Jasper, Microsoft Copilot or Google Gemini – can do amazing things: summarise mountains of information, plan a travel route, write code for apps or even translate texts. There’s just one problem: AI has no idea whether the result is actually meaningful and correct. And it can’t be relied on to check its veracity either.
Applied to (spoken language) interpreting, this means that AI also doesn’t know whether a speaker is joking. Or whether a speaker is being ironic. Or whether a person is deliberately expressing themselves in convoluted ways precisely because they want to obfuscate – or maybe simply because they don’t have a good command of the language. AI is also unable to summarise when too many words are spoken too quickly or when speakers repeat themselves endlessly, or complete a statement if the speaker doesn’t finish their sentence properly. That’s a skill that only we – human interpreters with human intelligence – have.
How does this matter to you? Well, if the presentations and remarks to be delivered during your event are not scripted down to the last detail, if there is any humour, sarcasm or wordplay, if some of the statements could be indirect or when cultural differences might come into play, then an AI tool will almost certainly stumble. In cases like this, you’re well advised to let human intelligence process human communication. Not to mention questions such as who can actually guarantee that what is processed by ‘the system’ is kept confidential? Or who can be held liable if ‘the system’ makes mistakes?
In September 2024, Germany’s national Association of Conference Interpreters (VKD) published a position paper (in German) that systematically addresses all these issues and more. Now of course you could be forgiven for thinking that the VKD would naturally argue ‘pro domo’ (its own case). Then again, you may want to take the paper – translated by an AI tool, if you insist –as an opportunity to examine your thoughts around this issue and then contrast it with what you want to achieve with your event to verify whether the potential pitfalls of AI could become a problem – in which case you might prefer to make a ‘pro homo’ decision. If you do, get in touch and ask us for a quote.