MIT Technology Review got to try out Astra in a private live demo last week. It was a great experience, but there’s a difference between a polished promotion and a live demo.
Astra uses Gemini 2.0’s built-in agent framework to answer questions and perform tasks through text, audio, images, and video, and calls existing Google apps like Search, Maps, and Lens as needed. . “It integrates some of the most powerful information retrieval systems of our time,” said Bibo Xu, product manager at Astra.
Gemini 2.0 and Astra now include Mariner, a new agent built on top of Gemini that can browse the web. Jules, a new coding assistant powered by Gemini. Gemini for Games is an experimental assistant that lets you chat and ask for tips while playing video games.
(And don’t forget that last week, Google DeepMind also announced Veo, a new video generation model, Imagen 3, a new version of its image generation model, and Willow, a new kind of chip for quantum computers. Phew. Meanwhile, CEO Demis Hassabis was in Sweden yesterday after winning the Nobel Prize.
Google DeepMind says that Gemini 2.0 is twice as fast as its previous version, Gemini 1.5, and that it is a large set of multiple-choice questions designed to test the capabilities of large-scale language models. It claims to outperform it on many standard benchmarks, including Pro. It spans a wide range of subjects, from mathematics and physics to health, psychology and philosophy.
But the margin between top-end models like Gemini 2.0 and models from rival labs like OpenAI and Anthropic is now slim. Recent advances in large-scale language models are less about how good they are and more about what you can do with them.
That’s where an agent comes in.
Try using Project Astra
Last week, I was led through an unmarked door on the upper floor of a building in London’s King’s Cross district to a room that had the feel of a secret project. The word “ASTRA” was emblazoned in giant letters all over the wall. Xu’s dog Charlie, the project’s de facto mascot, wandered among the desks where researchers and engineers were busy developing the products on which Google staked its future.
