Google Announces New AI Model: Gemini

Google’s Gemini, the largest scientific and research undertaking in the company's history, stands as the culmination of extensive collaborative efforts across diverse Google teams, including Google DeepMind and Google Research. According to the company, Gemini’s state-of-the-art capabilities will significantly enhance the way developers and companies build and scale with AI.  

Gemini is a multimodal AI model, demonstrating the capability to seamlessly comprehend and generate multimedia data, including text, image, audio, video, and programming languages. Notably, it distinguishes itself as an exceptionally flexible model, operating efficiently across various devices, from mobiles to expansive data centers. Available on Google AI Studio, Vertex AI, and Google Cloud, developers will be able to customize Gemini “with full data control and benefits from additional Google Cloud features for enterprise security, safety, privacy, and data governance and compliance,” reads a Google blog post.

“Its remarkable ability to extract insights from hundreds of thousands of documents through reading, filtering and understanding information will help deliver new breakthroughs at digital speeds in many fields from science to finance,” writes Demis Hassabis, CEO and Co-Founder, Google DeepMind, in a company’s blog post.

Gemini will also help improve the processing power of Bard, the company’s generative AI chatbot. Through Gemini updates, Google expects to enhance Bard’s capabilities in a wide range of tasks, surpassing GPT-4 models in skills related to reasoning, planning, math, and code generation. “This is the biggest single quality improvement of Bard since we [ha]ve launched it,” said Sissie Hsiao, Assistant and Bard VP and GM, Google, in a press briefing. 

“With a score of 90.0%, Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding), which uses a combination of 57 subjects such as math, physics, history, law, medicine, and ethics for testing both world knowledge and problem-solving abilities,” explained Hassabis on Google’s blog. 

To minimize the potential for the generation of biased and toxic content, Google has committed itself to designing safety classifiers to identify, label and expel content involving violence and negative stereotypes. Through this approach, the tech company expects to make Gemini safer and more inclusive. To achieve this, Google has partnered with Frontier Model Forum and AI Safety Fund to define best practices and set security benchmarks accessible for various organizations within the company’s website. 

Bard with Gemini Pro will be accessible in English in over 170 countries, showcasing advancements in comprehension, summarization, reasoning, idea generation, writing, and planning. Additionally, Gemini will also be integrated with Pixel 8 Pro, the first smartphone designed to run Gemini Nano, introducing new features such as summarization in the Recorder app and Smart Compose in Gboard.