Google expands AI mode to Japanese, Korean, Hindi

Google is expanding its AI-powered search feature, AI Mode, to Japanese, Korean, Hindi, Indonesian, and Brazilian Portuguese, after launching in English in May.

The company said the feature will be available globally in these new languages, though it was not yet accessible in Japanese at the time of announcement.

AI Mode, built on Google’s Gemini 2.5 model, first launched in the US, the UK, and India, and expanded to 180 countries last month, but remained English-only until now.

AI Mode allows users to ask complex questions and receive responses that draw on a wider range of data, including images and video transcripts.

🔗 Source: The Japan Times

🧠 Food for thought

Implications, context, and why it matters.

Multilingual AI search faces massive scale and technical complexity

Google’s expansion of AI Mode to five new languages represents a significant technical challenge given the platform’s enormous scale of 5.6 billion searches processed daily 2.
The company’s initial English-only rollout to 180 countries before adding language support demonstrates the complexity of developing AI systems that can handle diverse linguistic queries while maintaining response quality.
This measured rollout approach contrasts with Google’s traditional search dominance, where it holds over 90% market share 2, but trails significantly in AI chatbots with only 2.2% compared to ChatGPT’s 81% share 1.
The technical complexity becomes evident when considering that 15% of Google’s daily searches are unique queries 3, meaning AI systems must handle unprecedented questions across multiple languages simultaneously.

AI localization requires cultural understanding beyond simple translation

Google’s emphasis on “nuanced understanding of local information” reflects broader industry recognition that AI localization involves cultural sensitivity, not just language conversion 4.
Industry analysis shows that AI translation systems often struggle with cultural context and can perpetuate biases present in training data, requiring human oversight to ensure cultural appropriateness 5.
The complexity extends beyond language barriers, as effective AI localization must integrate local terminology, cultural references, and region-specific information needs that vary significantly across markets 6.
This challenge explains why Google developed custom versions of Gemini 2.5 for each language rather than using a single multilingual model, acknowledging that truly localized AI requires dedicated development for each linguistic market.