Google is not done, though. Five days before Christmas 2024, the Mountain View, California-based tech giant expanded Gemini’s Deep Research mode’s language support to 40.
For now, this feature is only available to Google One AI Premium members. It empowers users to leverage Gemini’s “long context capabilities” and “advanced reasoning” to generate credible research briefs, which members can export to Google Docs for editing and refinement.
Gemini’s Deep Research function is a multi-step process. It starts with creating a research plan and finding relevant information before re-searching credible resources to extract the most accurate information. This process cycles several times to create an “in-depth” research report.
Expanding Gemini’s language support to 40 can present unique challenges to “in-depth” research quality.
For instance, finding credible information in Swahili or Urdu can be more challenging than searching for authoritative sources in Spanish or French.
Summarizing the research findings in the original Deep Search language (i.e., Swahili or Urdu) can be intimidating without messing up grammatical rules, syntax, and other language attributes.
It is worth pointing out that translation isn’t always perfect due to contextual effects, perception biases, regional linguistic (dialects) differences, and more.
Experts see Google Gemini’s Deep Research feature as a game-changer, although much work is required to train the model in languages other than the major ones, such as English, French, German, Russian, and Italian.
Still, Google’s battle cry is loud and clear: authoritative knowledge should be available to EVERYONE regardless of language.