“So tell me, when will Machine Translation replace humans?”
We all hear this a lot. As we deal with a wide variety of folks — many of whom are not in the industry but are responsible for some aspect of localization at their company – this has become an expected question.
After experimenting with countless other answers, we’ve hit on a simple way to successfully address this question. We’re happy to share it. AND we’d look forward to hearing from you if you have any tweaks to it. While (of course!) it’s open to detailed dissection, it’s not intended to be the final answer or a PhD thesis. The win here is that we found a language everyone can comprehend. It enables us to make the point that needs to be made.
“Machine Translation will completely replace humans on the same day AI will replace your CEO”
Here’s the story you can tell that explains why this is the case.
- Picture a company with a simple organizational structure.
- Think about job functions you can envision being replaced by AI. For example, some roles might be under Sales and Finance. You can imagine AI replacing someone in Sales Operations whose job is to collate, report, and comment on this month’s pipeline. You can also picture a case where someone reporting data to the CFO could be replaced by AI. To what extent have chatbots already replaced first-line Support personnel? As advances in AI continue, more jobs might be up for AI replacement or augmentation. The org chart is here, and we highlight in yellow potential jobs that have or could be replaced by AI. Over time, the list will likely grow.
Functions that deal with more human-centered, creative, and strategic decisions — branding, public perception, market direction, etc. – will be the last to be replaced. Even the most forward-thinking of us has a hard time envisioning when the CEO can be replaced. Someone has got to oversee all those algorithms!
- Now that you’ve got the listener in a place they understand, replace the org chart with a “language org chart”. At the top of our org sits the Head of Globalization (HOG) who decides which content is done by human linguists (L) or Machine Translation (MT). We replace Direct Reports with sample content that falls under each of these categories.
- Start with the fun story that some companies already translate their most straightforward Support content with MT. So long as the source content is written in a simple, declarative manner, MT can make the translations understandable and actionable…which fulfills the requirement for many knowledge base (KB) articles. Other content across the org written in a straightforward, declarative, and simple way might also become a candidate for pure MT. Like the human org chart mentioned earlier, technology improvements will mean additional candidates for MT.
But what about content/translations that contain any semblance of nuance, teach complex concepts, or aim to elicit an emotional response? What about translations of bad English? Even when NLP/AI/MT become that good, someone will need to manage those algorithms to make sure they’re working properly…which is why we’ll always need linguists. Ergo, MT replaces linguists at the same time AI replaces the CEO.
Responses to this explanation have, so far, been excellent. People get it when we use this language. It took us a while, but this methodology now helps shorten the answer to this common (sometimes frustrating) question. Give it a try; let us know how it goes!