Machine translation takes a giant leap forward
For the past few decades, machine translation has not made any major advancements. Sure, the machines have gotten smarter and their suggestions have gotten better—but only incrementally.
Lilt was founded by some of the smartest engineers I've had the pleasure to be around. Their backgrounds in machine translation—from PHD's at Stanford to core members of the Google translate team—have given a whole new light to how translators might leverage a machine to translate faster and at higher quality.
They had a lot of great back end technology in place. I came in to help turn it into a commercially viable translation tool. We started with a web app and a couple core principles.
Firstly, translation pairs should no longer live side-by-side (like an excel document), but rather be stacked on top of each other. This helps eye movement around the screen and keeps translators engaged with both the source text and their translation.
Secondly, Lilt's high quality machine suggestions should not live off on a side panel somewhere, but rather right inline with the text they are suggesting—and they should be interactive. This is a revolutionary approach to teamwork between humans and machines that emphasizes the strenghts of each. Machines are great at instant recall (in this case of millions of possible word pairings). Humans are exceptional at the intuition to know which suggestions are most appropriate and when to say something entirely differently.
Thirdly, translation tools, like dictionaries and examples, should be available inside of the translation experience. When a translator has a question, Lilt should be there waiting with the answer—and a an easy way to use that answer.
Lastly, the web app should be a shining symbol of quality. In a world of translation tools that are outdated at best, Lilt needs to be friendly, clean and modern. I took a very consumer-centric approach to the visual design of the app. Colors are warm and vibrant. Typography is bold and clear. Unecessary tools are tucked away or removed altogether.
Lilt will be launching soon. Private beta translation professionals has been going very well and they continue to make iterative improvements based on user feedback.