Check out this month’s highlights to learn how Subway saved $1.5 million and 2 years of labor using Intento, along with details on the growing focus on under-resourced languages, why a cognitive AI engine is the next step in accessibility and inclusion, and more. Enjoy!
💪 What winning looks like with MT ROI: Subway
🔦 Spotlight on more data for under-resourced languages
🤖 Access 40+ stock and custom model providers, now including Globalese
🌎 Why cognitive AI is the next step in accessibility and inclusion
🍶 GlobalSaké interactive virtual event June 2nd
📚 Great reads from throughout the community you won’t want to miss
Machine translation ROI: What winning looks like
At the XTMLive 2022 conference last month in Silicon Valley, Carrie Fischer (Manager of Globalization Services, Subway) was joined by Mark Hjerpe (Head of Revenue, Intento) and Rory Sampair (Solutions Architect, XTM) to discuss the clear, measurable returns delivered by machine translation.
If you missed the talk, read our summary for key insights on how it’s possible to save a bunch of money and time while enhancing your MT program.
A greater focus on more under-resourced languages
Microsoft has added Faroese for text and document translation — becoming one of the few providers that support this niche language. Microsoft offers some of the most dynamic coverage on the Intento platform — having just added Basque and Galician as well. This comes as Google expands its support to 24 additional under-resourced languages, using monolingual data alone to train MT models.
Take look at the full list of Intento language options here.
Globalese joins the 40+ Intento stock and custom model providers
Inento users can now access the new Globalese MT engine in all TMS systems we have connectors for (15 and counting)!
You’ll get the best experience with custom models, and Globalese offers two levels of customization. When you need a generic engine trained on an extensive dataset, there’s Stock+ customization. If you have enough data for training, their Domain-adapted models focus on style and terminology. Globalese engine version 4.1.2 or higher also supports content with inline markups.
Why a cognitive AI engine is the next step in accessibility and inclusion
The billions of dollars expected to pour into AI development will expand global accessibility and broaden the scope of inclusion by:
- Breaking down language barriers that hinder self-expression
- Building MT that can handle more sensitive translations
- Open doors for populations that rely on tone, intent, context, and gestures
Check out a fresh take from industry leaders Grigory Sapunov (Intento) and Andrew Gershfeld (Flint Capital) here.
Join us at the next GlobalSaké interactive virtual event June 2nd
Along with virtual events you may have noticed that the GlobalSaké community has been active with planning LocWalks, prepping for an in-person event on June 9, and launching the Mentorship Program. Use code GSVIP30 for a 30% discount on your annual pass!
More updates from the AI/MT community
- Unlocking zero-resource MT to support new languages in Google Translate
- Introducing J1-Grande
- Democratizing access to large-scale models
- Tackling multiple tasks with a single visual language model
- The great language model scale off: Google’s PaLM
- Researchers gain new understanding from simple AI
- An empirical analysis of compute-optimal large language model training
- What are the most important African languages in emerging markets?