Google plans to introduce AI feature 'Jarvis' to Chrome, which will allow users to book flights and buy products in their browsers
Google is developing an AI agent called 'Jarvis' that will perform tasks such as 'collecting research results,' 'buying products,' and 'booking flights' in users' web browsers, according to a report from overseas media The Information.
Google Preps AI That Takes Over Computers — The Information
Google is reportedly developing a 'computer-using agent' AI system - The Verge
https://www.theverge.com/2024/10/26/24280431/google-project-jarvis-ai-system-computer-using-agent
'Project Jarvis' leak highlights Google Gemini 2.0's superpower
https://www.androidpolice.com/google-gemini-project-jarvis-ai-agent/
Report: Google preps 'Jarvis' AI agent that works in Chrome
https://9to5google.com/2024/10/26/google-jarvis-agent-chrome/
At its annual conference, Google I/O 2024 , held in May 2024, Google announced thatit would directly incorporate the multimodal AI Gemini Nano into the desktop version of Chrome , and in July Gemini Nano was incorporated into the beta version of Chrome 127. Google also mentioned a 'universal AI agent that will be useful in everyday life,' and said that some of the AI agent's functions may be incorporated into Gemini.
The Information recently reported that Google is working on a project codenamed 'Project Jarvis.' According to three people familiar with the project, Jarvis will be equipped with 'Gemini 2.0,' a future version of Google's multimodal AI ' Gemini ,' and will perform tasks such as gathering information, purchasing products, and buying airline tickets on a web browser.
Jarvis is tailored specifically for Chrome, and it continuously takes screenshots of the screen, interprets them, and clicks buttons and enters text based on the user's instructions. However, at the time of writing, it seems to be a little slow to respond, and the reason for this is that 'the model needs to think for a few seconds before performing each action.' Technology media 9to5Google pointed out that 'Perhaps Jarvis does not yet run on-device and requires the cloud.'
A preview version of Jarvis is reportedly scheduled to be released as early as December 2024, which lines up with rumors that
Technology media Android Police said, 'Google wants to offer Jarvis to a small number of users for testing initially, and doesn't expect it to be widely available when it's introduced. The December release timeline is also not set in stone, and as The Information points out, Google may choose not to show off Jarvis and its features by then.'
Google isn't the only company working on AI models like Jarvis that work with browsers. Microsoft is developing Copilot Vision , an AI feature that can converse with text and images in the browser, and AI company Anthropic has started public beta testing a feature called computer use, where the AI model Claude will control a PC.
In addition, Microsoft announced a feature called 'Recall' in May 2024 that periodically takes screenshots of users' PC operations, saves them in a database, and allows users to check the operation history later. Recall has been postponed due to privacy concerns, but a preview version has been available via Windows Insider since October.
Related Posts:
in Software, Web Service, Posted by log1h_ik