Gemini 2.0: Google’s bid to get closer to building Universal assistants

Google introduced the next iteration of its generative AI model Gemini, on Wednesday. The Mountain View company calls Gemini 2.0 its most capable model to date. Google says that, along with its superior multimodal capabilities, this model will bring it closer to its vision of a universal assistant through AI agents. 

“If Gemini 1.0 was about organizing and understanding information, Gemini 2.0 is about making it much more useful. I can’t wait to see what this next era brings,” Google chief Sundar Pichai said in a blog. 

He further added that Gemini 2.0’s advances are a result of Google’s decade-long investments in its ‘differentiated full-stack approach to AI innovation’. It is built on custom hardware like Google’s sixth-generation TPUs Trillium. TPUs powered 100% of Gemini 2.0 training and inference, and Trillium is now generally available to customers too.

To begin with, developers and trusted testers will be getting access to the model. The company plans to start incorporating Gemini 2.0 quickly in its products, starting with Gemini chatbot and Search. Pichai noted in his blog that Search has transformed the most with AI; the Google AI Overviews now reach 1 billion people, making it one of the most popular Search features. With Gemini 2.0, AI Overviews plans to tackle more complex topics and multi-step questions, including advanced math equations, multimodal queries, and coding. Limited testing has already begun.

To begin with, Google has rolled out Gemini 2.0 Flash experimental model, a workhorse model with low latency and enhanced performance. In addition to supporting multimodal inputs like images, video and audio, 2.0 Flash now supports multimodal output like natively generated images mixed with text and steerable text-to-speech (TTS) multilingual audio. Gemini users can also access a chat-optimized version of 2.0 Flash experimental.

Additionally, Google has also announced a fresh update to Project Astra, a research prototype exploring the future capabilities of a universal AI assistant. The company has announced Project Mariner, which explores the future of human-agent interaction, starting with building an agent for web browsers; and Jules, an AI-powered code agent that can help developers.