Google has introduced its latest AI model, Gemini AI, asserting its superior performance over OpenAI’s GPT3.5 model with confidence.
As per the official announcement, the American tech giant revealed that the new model will empower its chatbot Bard with enhanced capabilities in tasks like text comprehension, logical reasoning, planning, and various other aspects.
Gemini is set to be released in three versions: Nano, Pro, and Ultra. Google aims to deploy this new intelligent model across all smart platforms, ranging from phones to data centers.
The integration of Gemini into Google’s chatbot “Bard” will occur in two phases. The initial phase will commence with the introduction of “Gemini Pro” to Bard, starting on Thursday. It will initially support the English language in 170 countries worldwide, with expanded language support planned for the coming months.
The second phase involves the introduction of a more advanced version of Bard, named Bard Advanced, which will rely on the Gemini Ultra version, the most advanced among the three Gemini iterations.
Google offers diverse capabilities in the three Gemini versions. Gemini Nano, the lightweight and less capable version, is designed for smartphones, with Google Pixel 8 and Pixel 8 Pro being the first phones to run the new model.
Gemini Nano utilizes Google’s latest Tensor G3 processor, supporting various features on Google’s phones launched last October, such as the Summarize in Recorder feature and Smart Reply with the Gboard keyboard app.
Gemini Nano ensures the processing of user data remains local on their phones, without sending any information to Google servers, guaranteeing the speed of AI features without the need for an internet connection.
The audio summarization feature enables users to record meetings, lectures, or traditional voice recordings, then convert them into texts. Users can press a specific button within the Recorder app to leverage Google’s smart model “Gemini Nano” to summarize texts and convert them into clear key points.
While the smart reply feature relies on “Gemini Nano” for Pixel 8 phones, it will be supported within all instant messaging apps, starting with WhatsApp later this year, with more messaging apps supported next year.
By 2024, Google’s smart assistant, Assistant, will gain the advanced capabilities of Bard, limited to Google Pixel phones only.
The Pro version of Gemini supports Google’s chatbot Bard, providing advanced capabilities in text analysis, generation, coding, and handling different input forms such as texts, images, videos, and audio clips simultaneously.
Google emphasized Gemini Ultra’s advanced capabilities in accomplishing complex tasks, scoring 90% on the MMLU standard, equivalent to the performance of human experts in handling a mix of 57 complex cognitive topics.
Gemini Ultra’s focus on the MMLU standard during training and development enhanced its ability to carefully consider responses to challenging questions, refining its proficiency in providing accurate information rather than relying on first impressions.
Google announced that Gemini Ultra would not be available to the public before 2024. However, the company plans to allow a limited sample of Bard users on the upcoming advanced version, Bard Advanced, which will be launched for general users next year.
Leave a Reply