Google expresses optimism regarding the potential integration of its more advanced artificial intelligence models into smartphones within the coming year. The tech giant anticipates that its presently available Gemini large language model (LLM), which competes with Microsoft-supported OpenAI’s formidable GPT-4 AI model, will commence integration into devices as early as next year.
Currently, Google offers Gemini Nano, its most efficient model for on-device AI, across its Pixel devices and other compatible Android devices. Brian Rakowski, vice president of product management for Google’s Pixel unit, foresees the company’s most sophisticated large language models, currently accessible solely via remote data centers through an internet connection, becoming available directly on smartphones next year.
“There are smaller versions of our Gemini model on the cloud,” Rakowski informed CNBC. “There’s been quite a few breakthroughs to compress these models to get them to run on the device.” “Some have already been proven and some are being explored for some applications. It’d be great to have all the models on the 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 emphasized. “You can do a lot with these distilled small versions of the models on the 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 a 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 manner. Gemini Ultra stands as Google’s premier LLM, boasting an impressive 1.56 trillion parameters. By comparison, OpenAI’s GPT-4 comprises 1.76 trillion parameters.
Envisioning a smartphone ‘supercycle’
Smartphone manufacturers have long envisioned a “supercycle” in their industry, propelled by artificial intelligence, following a challenging period marked by a significant slowdown in device sales. In 2023, smartphone sales plummeted to 1.16 billion units, marking the lowest point for unit shipments in a decade.
Analysts are skeptical about the likelihood of a supercycle materializing in the next few years, citing insufficient market dynamism in terms of innovative features to entice users to upgrade from their aging smartphones.
“Unfortunately, we don’t expect that boom,” remarked Francisco Jeronimo, vice president of data and analytics at research firm IDC.
“The last supercycle we saw was between 2010 and 2015, where in five years the market grew five times from around 300 million smartphones per year to 1.5 billion.” Nevertheless, an increasing number of smartphone manufacturers are making substantial investments in AI, hoping to inject renewed excitement into mobile technology.
Companies like Humane, Rabbit, and China’s Meizu are betting on a smartphone future that deviates from the traditional mold, envisioning smaller, more compact devices that users can interact with via voice activation, akin to an Amazon Echo speaker but portable. Google has been making significant strides in AI to gain a competitive edge over rivals like OpenAI, the Microsoft-backed entity behind ChatGPT.
Google recently unveiled a major rebrand of Bard, its alternative to ChatGPT, introducing a revamped app and subscription options. Bard was rebranded as Gemini, aligning with the suite of AI models powering the chatbot. Android users now have access to a dedicated app for Gemini, while iPhone users can utilize Gemini within the Google app on iOS.
During Alphabet’s Jan. 30 earnings call, CEO Sundar Pichai underscored the company’s commitment to AI. Pichai envisions offering an AI agent capable of handling increasingly diverse tasks on behalf of users, including within Google Search, although he acknowledged that “a lot of execution lies ahead.” Likewise, chief executives at tech giants ranging from Microsoft to Amazon have reiterated their commitment to developing AI agents as productivity tools.
