Google is continuing its push into the AI scene with a couple of major announcements this week, starting with the introduction of their new AI model: Gemini. The Gemini model is Google’s ‘largest and most capable’ AI model ever, while also boasting about its flexibility.
The result of large scale collaborative work between various teams at Google, Gemini was built from the ground up and Gemini 1.0 will feature three different sizes: Gemini Ultra, Gemini Pro and Gemini Nano. The Gemini Ultra is their largest and most capable model of the three, perfect for highly complex tasks, while Gemini Pro is their best model for scaling across different tasks. Gemini Nano meanwhile is tuned more towards efficiency, making it a great choice for on-device tasks. This kind of flexibility allows Gemini the potential to be used across everything from data centers to your smartphone in your pocket.
“Now, we’re taking the next step on our journey with Gemini, our most capable and general model yet, with state-of-the-art performance across many leading benchmarks.
This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company. I’m genuinely excited for what’s ahead, and for the opportunities Gemini will unlock for people everywhere,” Sundar Pichai, Google CEO
Gemini’s first big use case? Well in Google’s own Bard generative AI chat tool of course. Specifically, Google is bringing a tuned version of Gemini Pro to Bard, and it is already available for all Bard users now. This gives Bard the ability for more advanced reasoning, planning, understanding and more. Google claims that Gemini Pro had outperformed OpenAI’s GPT 3.5 in six out of eight benchmarks. They add that they’re already seeing Bard become the most preferred free chatbot out there compared to their rivals in blind evaluations with third-party raters.
To show off their new Gemini Pro-powered Bard, Google also teamed up with YouTuber and educator Mark Rober, with Rober using the new Bard to work on creating the best, most accurate paper airplane possible. You can check that out below:
On top of that, Gemini Ultra will be coming to Bard too, with Google teasing a new Bard Advanced due to arrive sometime next year. It will feature a ‘cutting edge AI’ experience, using Gemini Ultra’s ability to understand and act on different types of information such as text, images, audio, video and code. At the moment though, Google is working on ensuring Bard Advanced and Gemini Ultra is a safe and trusted model, and so are currently completing extensive safety checks on it and will also put it through a trusted tester program before opening it up to the public.
For more about Gemini, you can check out Google’s blogposts on their new AI model here.