Right now, Gemini Pro and Gemini Ultra exist in large data centers and can only access them through the web while Gemini Nano is small enough to be squeezed into a phone. However, according to at least one high-level Google executive, this may not be the case for long. A VP at Google believes that the company’s largest language models could make it to phones sometime next year.
In case you don’t know, Gemini Nano is currently on smartphones such as the Google Pixel 8 Pro and the Galaxy S24 series. Recently, several MediaTek Dimensity processes gained compatibility with Gemini Nano. Because of this, people are able to use on-device generative AI like text creation and translation. More devices will gain on-device generative AI as time goes on.
Google execs believe that some of its largest language models could be on phones next year
Large language models are gigantic in size and much too big to be able to fit on mobile hardware. However, Brian Rakowski, vice president of product management for Google’s Pixel unit, begs to differ. He believes that some of Google’s larger language models will be able to fit on phones at some point next year. 2024 has just gotten started, so there’s still plenty of time between now and next year.
“Gemini Nano is performing at a level that our online models were at less than one year ago,” he told CNBC. “You can do a lot with these distilled small versions of the models on device.”
This makes sense, as we’ve been able to see some pretty powerful generative AI being shown off by our smartphones. A great example of this would be the Samsung Galaxy AI tools. There are several tools that generate text and translate spoken words with great proficiency.
Technology is only going to advance as more companies pour billions of dollars into AI technology. So, there’s no telling what sort of AI we’ll see on our phones next year. For all we know, AI models as powerful as Gemini Ultra could be landing on the Pixel 10 series
2024-02-26 15:05:36