Google expects its more advanced large language models to arrive on Android phones next year

Google is bullish on the prospect of its more advanced artificial intelligence models coming to smartphones in the next year.  

The internet giant expects that its currently available Gemini large language model (LLM), which competes with Microsoft-backed OpenAI's most powerful GPT-4 AI model, will begin to get embedded into devices starting next year.  

Google already offers Gemini Nano, which is the company's most efficient model for "on-device" AI, across its Pixel devices and on all other capable Android devices. 

Brian Rakowski, vice president of product management for Google's Pixel unit, said that he expects the company's most advanced large language models, which are currently only accessible via remote data centers via an internet connection, to begin arriving on smartphones directly next year. 

"There are smaller versions of our Gemini model on the cloud," Rakowski told CNBC. "There's been quite a few breakthroughs to compress these models to get them to run on device." 

"Some have already been proven and some are being explored for some applications. It'd be great having all models on device. It still has wondrous applications." 

"Gemini Nano is performing at a level that our online models were at less than one year ago," Rakowski added. "You can do a lot with these distilled small versions of the models on device." 

"If you just follow that trajectory, some of that stuff we thought we would have had to go to the cloud for next year will be on device, which is pretty exciting, which is instantaneous without requiring a connection or subscription." 

Large language models, or LLMs, are AI models capable of understanding and generating language in a humanlike way. Gemini Ultra is Google's top LLM, clocking in at a