Artificial Intelligence & Machine Learning

How Slate can work on a desktop computer when Google, Microsoft and other machine translation systems run on huge computer farms with advanced artificial intelligence that require massive databases?

Slate uses the artificial intelligence and machine learning technologies that dozens of companies used to build cloud-based services. Cloud-based services optimize the technologies for many users while we optimize Slate for your desktop. Modern computers have more computing power than huge server farms from 10 years ago. With as few as 80,000 segment pairs in your translation memories, Slate learns your personal translation style and creates the databases it needs.

This table demonstrates the differences between the users that Slate is designed for compared to the others.

CategoryGoogle Translate [1]Slate Starter Edition
per million characters
LocationAround the worldYour computer
Total users500 million 1
Words per day100 billion200,000
Language pairs10,5061,056
Live language pairs10,5061
Contributors9.5 million1
Contributions90 million140,000
Edit-distance zero[2]5% – 9% [3]25% – 40% [4]
Core technologyGoogle NMTMoses SMT

[1] Ten Years of Google Translate
[2] Percentage of MT segments that exactly match a desired reference segment.
[3] Derived from table in Machine vs Human Translation
[4] Actual customer reports. Example: Practical MT Evaluation For Translators