The ability to quickly and automatically translate anything you see through a web service is a powerful one, yet a tolerable version of a foreign article, menu or street sign. Should not this amazing tool be put to better use?
By combining the expertise of human translators with the speed and versatility of automated ones, you get the best of both worlds – and potentially a major business opportunity.
The problem with machine translation, when you really get down to it, is that it's bad. Sure, it does not mistake "tomato" for "potato," but it can not be trusted to do anything beyond the literal meaning of a series of words. In many cases that's all you need ̵
This is more than a convenience issue; for many, language provides serious professional and personal barriers.
"Co-founder and CEO Spence Green," said Lilt ;
Much of this information is not amenable to machine translation, he explained. Imagine if you were to do any work in a country where immigration law is not available in your language.
"Books, legal information, voting materials … when quality is required, you Need a human in the loop, "he said.
Google, where he interned in 2011 the systems do.
His realization, which he pursued with co-founder John DeNero, which was that machine learning system worked as a tool for translation, but as a tool for translators .
The basic idea of Lilt's tool is that the system provides translations for the next sentence or paragraph, as a reference for structure, tense , idiom and so on that the translator can consult and, at least, potentially, work faster and better. Lilt claims a 5x increase in words translated, and the results are as good as or better than a human translation.
"We published papers – we knew the technology worked. We've done through. " " We went through Green said.
Stay in academic research, get a grant and open-source it? "The money child of dried up," Green explained: money was lavishly allocated after 9/11 with the idea of improving intelligence and communication, but a decade later the sense of urgency had departed, and with [a lot of the grant cash.] "We knew the technology was inevitable," he said.
Interestingly, a major change in language translation took place around the time they were really getting to work on it. Statistical neural network system gave way to attention-based ones; These are sentences that apply to each other in a structured way. They have had to reinvent their core translation system, but it ultimately for the better.
"These systems have much better fluency – they're just a better model of language." Second, they learn much faster; You need fewer updates to adapt to a domain, "Green said.
Of course, you can not just sprint into the midst of the translation business, which spans publishing, real-time stuff, technical documents and a dozen other verticals, and say "here, use AI!"
"There's enormous structural resistance in the industry to automating in any real way," Green said.
"We tried several business models before we found one that works. "Okay, this human-in-the-loop method is the fundamental way to solve this problem, let's just build a company around that." So we're vertically integrated, we work
A faster method that does not apply to translation
Think about it like this: if 196 if speak puts products puts products products products in products in products in products in products in in in in in in in in in in in in in in in in in in in in a task that's essentially never done.
"We work with Zendesk, Snap, Sprinkler … we just take over the whole localization workflow for you. That helps with international go to market. "Said Green. If a company's budget and process before using it is limited to 5 or 6 new markets in a given period, that could double or triple for the same price and staff, depending on efficiency gains.
Right now they're working on acquiring customers, naturally. "In Q4 last year, we built our first sales team," Green admitted. But initial work has always been heartening, since they have "more idiosyncratic language needs" and a large volume of text. The 29 languages Lilt supports right now will be 43 by the end of the year. A proofreading feature is in the works to improve the efficiency of editors as well as translators.
Lilt. Academics are both a crucial source of translators and language experts and a major market.
Green's pet peeve seems to be the bright researchers that are making a great deal of work on boring consumer stuff: "Tech companies AI and Robotics.
Finally, Green said, "It's a great hope that we can close this circle and get into book translation as we go on. It's the lucrative work but it's the third part of the vision.
Although it may start out as support documents for and random government contracts, the types of content and markets are amenable to Lilt's type of human-in-the-loop process And a future where AI and people work in cooperation is certainly more reassuring than one where humans are replaced. With translation at least, the human touch is nowhere near ready to be retired.