Friday, September 20, 2024

Google Finally Explains How Gemini Became The Name For All Of Its AI Features

Must read

Google’s largest and most capable Artificial Intelligence (AI) model was first mentioned at I/O 2023. Since then, Gemini has been defining all of Google’s AI features. The company recently announced major feature updates for Gemini at its I/O event and explained how it will shape the future. After all the AI announcements, the company felt it was high time they explained how Google Gemini got its name.

Google officially explained the origins of Gemini, after it became the user-facing name for its AI features

Gemini is a Latin word that means “twins” in its literal sense. In astronomy, it is the northern constellation and is linked to the brightest stars, Pollux and Castor. According to Google’s DeepMind team, the project was originally supposed to be named Titan, which is the name for Saturn’s largest moon. Although the technical lead, Jeff Dean, did not agree with the scheme, it did inspire him for the name.

He went on to highlight the decade of breakthroughs in AI research that DeepMind and the Brain team achieved together. Some of the salient works include deep learning at scale, the Transformer architecture for all LLMs, deep reinforcement, and AlphaGo. In April 2023, Google DeepMind was born, which brought these two teams under one roof to work on multimodal AI models.

The Gemini effort came about because we wanted to bring our teams working on language modeling closer together,” Jeff says. “I felt the twins aspect of the name ‘Gemini’ was a great fit. The twins here are the folks in the legacy Brain team and the legacy DeepMind team, who started to work together on this ambitious multimodal model project.”

If we look at past trends for Google’s product name inspirations, we see that Google tends to opt for astronomical themes. Considering the dual personality the Gemini zodiac sign represents, the twin aspect of the word Gemini does justice to the work of the Brain and DeepMind team for the project.

Another aspect of the inspiration is also space-related. The AI model was named after NASA’s early moonshot program, Project Gemini, which lasted from 1965 to 1968. The project helped test the technology ahead of Apollo’s moon landing. The name did end up striking a chord with the team, as they could relate the efforts put forth for training LLMS to the ambition and work behind the Apollo program.

Project Gemini’s significance to the success of the Apollo program resonated with the team. At one point, Jeff proposed the name “Gemini” in a comment in a Doc the team was working on, and it stuck. “I was immediately sold on the name because the monumental effort of training LLMs resonated with the spirit of launching rockets,” says Oriol Vinyals, Gemini’s co-technical lead.

While the company explained in detail how it stumbled upon the name Gemini, we see Google’s vigorous efforts to promote its multimodal model and capabilities. As part of its Gemini upgrade, Google announced a handful of new AI features which will be available with Android 15 later this year on all compatible Pixel devices.

Share this story

Facebook

Twitter

Latest article