Roblox has built an artificial intelligence (AI) tool that can translate supporting languages in chats in almost real-time. There is a latency of roughly about 100 milliseconds, claims the company. The online gaming platform, which has more than 70 million daily active users, created a native large language model (LLM) to build the tool that can translate any combination of 16 languages. The foundational model behind the feature puts a heavy emphasis on accuracy, fluency, as well as the context of the rest of the sentence to find the best translation.
Announcing the new feature in a post, Roblox Chief Technology Officer Daniel Sturman said, “Using AI to automate real-time translations in text chat removes language barriers and brings more people together, no matter where they live in the world.” The tool currently supports English, Chinese (Simplified and Traditional), French, German, Indonesian, Italian, Japanese, Korean, Portuguese, Russian, Spanish, Thai, Turkish, Polish, and Vietnamese.
The feature activates on its own in chat boxes. When multiple users type in different languages, the AI will automatically translate them in real time to the user’s default language. A translation icon is placed in front of the sentence, clicking which turns the language back to the original.
During the process of creating the AI tool, the online gaming platform came across a strange challenge and found that while translating one language to another was simple, working in a combination of 16 languages was rather tricky. So, instead of building 256 different models (16 x 16), the company took a novel approach. Sturman explained that a unified, transformer-based architecture was used for the LLM where all the translation is controlled by the source sentence and the target language. “This is like having multiple translation apps, each specializing in a group of similar languages, all available with a single interface,” he added.
This system allows for some unique features. Apart from keeping an increased focus on accuracy, fluency, and context, the AI tool can also translate sentences where a combination of different languages has been used. Further, human evaluators were used to teach the model modern slang and trending phrases in each of the 16 languages. The process will be repeated continually to keep the LLM updated.
Additionally, the company also used a ‘back translation’ technique for less common translation pairs such as French to Thai. Here, Roblox first translated content and then translated it again to the original language. Then the source material was compared to the back-translated version for inaccuracies. Then, labelled data would be used to expand the amount of translation data for the model to learn.
Not stopping here, Sturman revealed that Roblox was now exploring automatic voice chat translations for its users.