Google has introduced Gemini, a new artificial intelligence model that it says is superior to previous AI models in terms of comprehension, summarization, reasoning, coding, and planning. Pro, Ultra, and Nano are the three versions of Gemini available. The Ultra version is scheduled for release in the first part of 2019; the Pro version is currently accessible.
Google, a direct rival of ChatGPT, has incorporated the new Gemini Pro with its chatbot Bard. Users can communicate with the Gemini-powered Bard through text. Additional modalities will be supported “soon,” according to Google. The new update is only available in English and is accessible in 170 countries and territories. Indian users are also receiving the latest version. Gemini’s abilities can be accessed through the Google Bard chatbot.
Gemini, according to Google, is the most powerful AI model ever created. The chatbot Bard is one tool we can use to verify Google’s claims. The fact that Gemini was designed from the ground up to be a multimodal tool sets it apart from other well-known AI models. A few examples of the new AI model’s ease of interaction across text, audio, and video were released by Google. In order to improve the interaction’s human-like qualities, it can also blend these modalities.
Google has been putting Gemini, a new AI model, to the test on a range of assignments. On thirty of the thirty-two standard academic tests used in AI research and development, Gemini Ultra, the most sophisticated version of Gemini, has outperformed the current best results. On an exam known as MMLU (massive multitask language understanding), Gemini Ultra received a score of 90.0%. 57 subjects, including maths, physics, history, law, medicine, and ethics, are used in this exam to assess knowledge and problem-solving skills. For the first time, an AI model outperformed human experts on this particular test.
Popular programming languages like Python, Java, C++, and Go can be understood and explained, and high-quality code can be generated by the first version of Gemini. This, according to Google, makes it one of the world’s best AI models for coding. Gemini Ultra passed multiple coding tests with flying colours. These include the industry-standard HumanEval for evaluating coding task performance and the author-created Natural2Code test, which replaces web-based information with author-created sources.
Additionally, more sophisticated coding systems can be powered by Gemini. AlphaCode, the first AI system capable of producing code that can compete in programming contests, was introduced by Google two years ago. Google developed AlphaCode 2, a more sophisticated code generation system, using a customised version of Gemini. Competitive programming problems involving advanced mathematics and theoretical computer science are well-solved by this system.
See also: Mount Merapi: 22 people have died as a result of Indonesia’s volcano