Google has announced the expansion of Gemini's in-depth research mode to 40 more languages, allowing users of its AI premium plan to access a more comprehensive research assistant. The feature, launched earlier this month, employs a multi-step approach to create a research plan, find relevant information, and extract knowledge to generate a report.

The expanded language support includes Arabic, Bengali, Chinese, Danish, French, German, Gujarati, Hindi, Indonesian, Italian, Japanese, Kannada, Korean, Malayalam, Marathi, Polish, Portuguese, Swahili, Spanish, Tamil, Telugu, Thai, Ukrainian, and Urdu. This development marks a significant milestone in Google's efforts to make its AI-powered research tool more accessible to a broader global audience.

However, the company faces a significant challenge in ensuring the reliability of sources in each language and summarizing the information accurately without compromising grammar. In an earlier conversation with TechCrunch, HyunJeong Choe, director of engineering for the Gemini app, acknowledged that while the company trains its model using clean data and trustworthy sources, AI overviews in native languages like Hindi tend to have inaccuracies in summaries.

To address this issue, Google relies on native sources of data and utilizes Google search on the backend to validate the information. The company also conducts evaluations and fact-checks in native language data before rolling out the model. Choe emphasized that factuality or getting correct information is a well-known research problem for generative AI in general, and the company is focusing on training the model to use the information in the right way.

Jules Walter, product lead for international markets for the Gemini app, revealed that the company has testing programs in place to ensure quality checks from native perspectives. This involves generating data to train models, which are then reviewed by local teams to ensure accuracy and relevance.

In a related development, TechCrunch reported earlier this week that a contracting firm working to improve Gemini by rating responses had received guidelines from Google stating that contractors were no longer allowed to skip prompt responses, regardless of their expertise. A Google spokesperson clarified that contractors not only rate answers for content but also evaluate style, format, and other factors.

The expansion of Gemini's in-depth research mode to 40 more languages marks a significant step forward in Google's AI research initiatives. As the company continues to refine its model and address the challenges associated with generating accurate summaries in native languages, the implications for global research and knowledge sharing are substantial.

With this development, Google is poised to further democratize access to AI-driven research tools, enabling users worldwide to tap into the power of machine learning and artificial intelligence. As the technology continues to evolve, it will be interesting to observe how Google addresses the complexities of language and cultural nuances to provide a seamless user experience across diverse markets.