Google is building its Gemini Nano AI model into Chrome on the desktop

Credits:  Frederic Lardinois  And Https://Techcrunch.Com/ 3:36 PM PDT • May 14, 2024

Image Credits: TechCrunch

At the Google I/O 2024 developer conference on Tuesday, Google announced that it is building Gemini Nano, the smallest of its AI models, directly into the Chrome desktop client, starting with Chrome 126.

This, the company says, will enable developers to use the on-device model to power their own AI features. Google itself plans to use this new capability to power features like the existing “help me write” tool from Workspace Lab in Gmail, for example.

The company says it’s the recent work on WebGPU and WASM support in Chrome that enables these models to run at a reasonable speed on a wide set of hardware.

In a briefing ahead of Tuesday’s announcement, Jon Dahlke, Google’s director of product management for Chrome, noted that Google was in talks with other browser vendors to enable this — or a similar feature — in their browsers, too.

“We have started to engage with other browsers and will be opening up an early preview program for developers,” Dahlke wrote in Tuesday’s announcement. “With webGPU, WASM, and Gemini built into Chrome, we believe the web IS AI-ready. “

I doubt that most Chrome competitors would want to solely bet on Google’s AI models, though. What makes more sense is to enable browsers — and developers — to run the model of their choice. Google would clearly opt to run Gemini for its applications, but these models are small enough to allow developers to choose whichever one they want for their applications.

Google’s bet, however, is to enable a number of high-level APIs in Chrome to translate, caption and transcribe text in the browser using its Gemini models.

“To deliver this feature, we fine-tuned our most efficient version of Gemini and optimized Chrome,” Dahlke said during today’s developer keynote at I/O. “Now we want to give you access to Gemini models in Chrome. Our vision to give you the most power AI models in Chrome to reach billions of users without having to worry about prompt engineering, fine-tuning, capacity and cost. All you have to do is call a few high-level APIs – translate, caption, transcribe. This is a big shift for the web and we want to get it right.”

For developers, Google is now also using the built-in Gemini Nano model to power some new features in the Chrome DevTools Console. Thanks to this, the Chrome dev tools can now explain errors and provide debugging solutions right in the console.

There's An AI For That (TAAFT) - The #1 AI Aggregator.

The Largest Database Of 12,514 AIs Available For Over 15,231 Tasks. Use Our Smart AI Search To Find The Best AI Tools For Any Use Case. Sign Up Here To Get The AI Tool That Helps You Find Other AI Tools For Free. Follow Us For The Latest AI Trends, Insights, And Updates. Let's Unlock The Potential Of AI Together!





Comments

Post a Comment

Popular posts from this blog

Google mentioned ‘AI’ 120+ times during its I/O keynote

Image
Credits:   Brian Heater   And Https://Techcrunch.Com/    2:37 PM PDT • May 14, 2024 Image Credits:  Google (screenshot by TechCrunch) It ran 110 minutes, but Google managed to reference AI a whopping 121 times during  Google I/O 2024  (by its own count). CEO Sundar Pichai referenced the figure to wrap up the presentation, cheekily stating that the company was doing the “hard work” of counting for us. No surprise, of course, that the topic took center stage, but even so, it’s an impressive figure for a presentation that ran roughly 110 minutes, all told. Gemini and its various iterations stole the spotlight (after preshow warmup act Marc Rebillet was finished with it). Google is incorporating the LLM-based platform into virtually every one of its offerings, including Android, Search and Gmail. The tech giant had a hard act to follow, with  OpenAI’s last GPT announcement  sucking a fair bit of air out of the room at Shoreline Amphitheater ...
Read more

LearnLM is Google’s new family of AI models for education

Image
Google says it has developed a new family of generative AI models “fine-tuned” for learning:  LearnLM . Credits: Kyle Wiggers And Https://Techcrunch.Com/ 1:15 PM PDT • May 14, 2024 Image Credits:  Google A collaboration between Google’s DeepMind AI research division and Google Research, LearnLM models — built on top of Google’s Gemini models — are designed to “conversationally” tutor students on a range of subjects, Google says. LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom. “LearnLM is grounded in educational research, making learning experiences more personal and engaging,” J ames Manyika, SVP of research, technology and society at Google Research, said onstage during a keynote at  Google’s I/O 2024 developer conference on Tuesday .  Google says that, through a pilot program in Google Classroom, it’s working with educators to see how LearnLM might simplify and improve the ...
Read more

OpenAI inks deal to train AI on Reddit data

Image
OpenAI has  reached a deal  with Reddit to use the social news site’s data for training AI models. Kyle Wiggers  3:04 PM PDT • May 16, 2024 Image Credits:   TechCrunch In a   blog post   on OpenAI’s press relations site, the company said that the Reddit partnership will provide it access to “real-time, structured and unique content” — e.g. posts and replies — from Reddit, allowing its tools and models to “better understand and showcase” that content. Reddit content will be incorporated into   ChatGPT , OpenAI’s popular conversational AI, and the companies will work together to bring unspecified new “AI-powered features” to both Reddit users and moderators. OpenAI will also become a Reddit advertising partner. “Reddit will be building on OpenAI’s platform of AI models to bring its powerful vision to life,” OpenAI wrote in the post. “Using LLMs, ML, and AI allow Reddit to improve the user experience for everyone.” OpenAI has several similar licensing dea...
Read more