Gemini, Google’s generative AI, continues to evolve. Today, it already rivals ChatGPT and is capable of understanding many formats, in addition to text, thanks to its multimodal capabilities. But the next step in the development of Gemini will allow Google to offer “AI agents” that will save Chrome users even more time.
In a press release published this week, Google presents Gemini 2.0. Available in test version, this one is still in development. And one of its main new features is that it will allow the firm to offer AI agents which will be capable of carrying out actions on the internet in place of the user.
Building on the new capabilities of Gemini 2.0, Google has developed a prototype called Mariner, which is capable of taking control of Google Chrome to perform online actions for the user, with the aim of saving time. In the video below, Google gives Mariner a list of businesses and asks the AI to search for their contacts. Mariner then takes control of Google Chrome and performs the searches for the user, ultimately producing the requested contact list. During the entire process, Mariner details in real time the actions he performs as well as his reasoning.
“Agents”, the future of artificial intelligence
Another video posted by Google suggests that, thanks to the AI agent, it will be possible to entrust a search to Gemini, for a specific type of product, then add the desired product to an e-commerce site.
But at the moment, it is not yet a finished product. “Project Mariner is an early research prototype built with Gemini 2.0 that explores the future of human-agent interaction, starting with your browser. As a research prototype, it is able to understand and reason through information from your browser screen, including pixels and web elements such as text, code, images and forms, and then to use this information via an experimental Chrome extension to perform tasks for you”explains the firm.
However, she admits that, for the moment, AI may produce incorrect results or take too long to complete certain tasks. But Google is confident that its AI performance will improve over time. For the moment, the Mariner project is only offered to a group of testers, via a Chrome extension. Alternatively, Google has also developed similar agents for developers to automate tasks in coders’ workflow.
Google continues to tease the Astra project
Google also continues to develop the Astra project, which it presented at its I/O conference in May. As a reminder, Astra is an assistant that uses the camera of a smartphone or connected glasses to observe the user’s environment and answer questions related to this environment. The new version of this assistant is based on Gemini 2.0, which improves performance. For example, thanks to Gemini 2.0, Astra improves its language skills.
The future assistant can also work in conjunction with Google Search, Google Lens or Google Maps, to be even more useful. And thanks to the performance of Gemini 2.0, it also has lower latency. Google explains that it continues to develop Astra in order to offer this visual assistant on the Gemini mobile application, and on connected glasses.
You can test Gemini 2.0 Flash today
At the moment, we don’t know when Google will allow us to use its AI agents. And it does not indicate when the Astra project will be available. On the other hand, if you want a taste of the firm’s next generation of AI, you can start using Gemini 2.0 Flash today. Just go to the Google chatbot and select this new model.
“In addition to supporting multimodal inputs such as images, video, and audio, Flash 2.0 now supports multimodal outputs such as natively generated images mixed with text and multilingual audio that can be steered text-to-speech (TTS). It can also natively call tools like Google Search, code execution as well as third-party user-defined functions”indicates the firm.
- Google presents Gemini 2.0, the next evolution of its generative AI
- Thanks to its new capabilities, this AI allows Google to develop agents who not only answer questions, but who can also take control of Google Chrome to carry out actions
- Gemini 2.0 also allows Google to improve the Astra project
