In a blog post, Google described a strategic shift in how its AI handles language. The company says it is moving beyond conventional text translation and aiming to build models that understand the world's languages as they are actually used. This points to a broader goal of preserving the nuance of living languages rather than settling for word-for-word conversion.
The post frames this work as "AI for everyone in every language," suggesting that language understanding is essential to making AI useful globally. The reference to "rich, living languages" signals an interest in dialects, cultural expression, and the ever-changing nature of speech and writing.
Because this is a single source, there are no differing views to compare. The statement is a clear, high-level commitment to a new direction. It remains to be seen how this ambition will show up in products or research, but the focus on languages "exactly as they are expressed" marks a notable shift from the typical emphasis on standardized translation output.