Alphabet, the parent company behind Google, announced its groundbreaking Gemina AI model, which is expected to be a direct competitor to ChatGPT.
Gemini is a significant upgrade over Alphabet’s current Bard AI chatbot, which worked as a standalone app and saw mixed reviews upon release. One of Bard’s newest updates integrates parts of the Gemini technology to improve its responses.
“This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company,” Alphabet CEO Sundar Pichai states in a blog post.
Since the launch of ChatGPT, tech companies around the world have been racing to launch competitive AI products. There was even a rumor that Google initiated a “code red” alert to reassign team members to build AI models (though Pichai denies this).
Google is also releasing its newest tensor processing unit (TPU) used to train AI models. Dubbed, The Cloud TPU v5p, the new processor can train models three times faster than its previous model.
At the same time, the new processors address the concerns regarding how much computing power to train AI models.
According to the team that built the model, Google’s DeepMind AI unit, the new language model is the largest one yet. Despite its size, it’s actually cheaper to provide consumers with than Google’s prior models.
“It’s not just more capable, it’s also far more efficient,” Eli Collins, VP of Product with DeepMind states. “The latest model still requires a substantial amount of computing power to train, but Google is improving on its process.”
The new chips are available
There will be three tiers of Gemini. The mid-tier model, Gemini Pro is already being used to power Bard, while Gemini Ultra is expected to be powering Bard-Advanced by 2024. The versions will range from supporting entire data centers to a single mobile device.
The new chips will be available for preview on Wednesday, says Google.