Google Unveils Gemini: A Multimodal AI Model Outperforming OpenAI's GPT-4
Summary:
Google has unveiled Gemini, its latest artificial intelligence model, claiming superiority over OpenAI's GPT-4. The multimodal model can combine different types of information, and perform mathematical and specialized coding tasks. It achieved “state-of-the-art performance” across 30 out of 32 evaluations. It scored 90% on the massive multitask language understanding test, surpassing human expert performance. Incorporating advanced programming capabilities, Gemini is the most potent free chatbot, outperforming GPT-3.5 in six out of eight benchmarks. The model is also integrated into Google’s flagship phone, Pixel 8 Pro with forthcoming rollouts in more Google products.
Google has officially unveiled its most recent artificial intelligence model, Gemini, which is reportedly superior to OpenAI's GPT-4 model. On December 6th, Google's CEO Sundar Pichai and Demis Hassabis, CEO and co-founder of Google DeepMind, introduced Gemini through a company blog post. This new AI model, which comes in various sizes and use cases (Ultra, Pro, Nano), is multimodal, capable of integrating diverse information types. Furthermore, with its proficiency in mathematics and specialized coding, Gemini goes a step beyond OpenAI's GPT-4, which lacks mathematical abilities.
Google is proud to report that their Ultra version exhibits “state-of-the-art performance” in 30 out of the 32 academic benchmarks used to evaluate large language model (LLM) development. It also achieves a score of 90% on a massive multitask language understanding (MMLU) test, surpassing human expert-level performance, as per Google's claims.
Jeff Dean, Google’s Chief Scientist, praised Gemini Ultra as the first model “to accomplish human-expert performance on MMLU across 57 subjects while achieving a score of over 90%.”
Gemini's design enables it to move seamlessly through text, images, audio, and video, putting it ahead of its competitors. “Gemini was conceived to be multimodal from inception, not simply a text model with visual and audio encoders added afterwards,” explained Dean.
In terms of advanced programming abilities, Gemini can generate high-quality code with AlphaCode 2, a sophisticated code creation system. It can also tackle complex programming problems and work in synergy with developers. According to AI expert Rowan Cheung, Gemini Pro outperformed GPT-3.5 in six of eight benchmarks, meaning it is currently the most potent freely available chatbot.
If interested in experiencing this new AI model, a refined version of Gemini Pro has been incorporated into Google’s version of ChatGPT, known as Bard. Google spokespersons proclaimed, "This is the most significant upgrade for Bard since its inception."
An exciting feature of Gemini involves its imminent appearance on Google’s flagship phone, the Pixel 8 Pro. "Pixel 8 Pro will be the debut smartphone to run Gemini Nano, fuelling features like 'Summarize' in the Recorder app and 'Smart Reply' in Gboard, implementing in WhatsApp initially," Google announced.
The tech giant has already begun to experiment with Gemini's potential to reshape its dominant web search engine into a more generative experience. Google initially teased Gemini earlier this year, boasting about its potential and claiming it would outperform ChatGPT.
Published At
12/7/2023 6:26:23 AM
Disclaimer: Algoine does not endorse any content or product on this page. Readers should conduct their own research before taking any actions related to the asset, company, or any information in this article and assume full responsibility for their decisions. This article should not be considered as investment advice. Our news is prepared with AI support.
Do you suspect this content may be misleading, incomplete, or inappropriate in any way, requiring modification or removal?
We appreciate your report.