Google CEO Sundar Pichai predicts profit as Gemini 1.5 utilizes AI to swiftly evaluate vast amounts of content

Google’s new artificial intelligence model can read Leo Tolstoy’s massive novel “War and Peace,” which spans half a million words and exceeds 1,000 pages, in just a minute.

The Alphabet-owned company on Thursday introduced the Gemini 1.5, an update to the AI model series it announced two months ago. With a single prompt, a user can ask Google’s AI to analyze significantly more content than before — 30,000 lines of code, 11 hours of audio, a full hour of video, Google said.

It’s for the Pro, the mid-size model in the family. In concrete terms, the Pro can process data known as a million tokens – five times more than Anthropic, perhaps a competitor in this regard, processes in a survey of its Claude 2.1 technology.

Reuters was unable to independently test the claims.

Google’s increasingly greedy AI, with ChatGPT creator OpenAI and other releases, reflects the rapid arms race underway in Silicon Valley to build the most powerful — and marketable — technology yet.

In an interview with Reuters, Alphabet CEO Sundar Pichai heralded the advance as one of several “breakthroughs” that would fuel his company’s myriad ventures. In essence, he said, the mainstay of the search industry enabled a new way of doing research.

“You give people a dramatically broader view to ask questions about the world.”

One example that Google discussed internally is how a filmmaker can ask an AI to rate a rough movie like a critic would, Pichai said. “This is just one use case we talked about as a team, but the sky’s the limit,” he said.

Another: polling multiple companies’ financial reports at once, Pichai said. Press demos showed how the AI was able to extract information from a 44-minute video in about 59 seconds, as well as multimodal prompts where the user asked the AI to respond to a combination of text and images.

In addition to bringing updated technology to Google search and YouTube, Alphabet wants Gemini 1.5 to attract customers to its cloud unit. Competing with larger rival and OpenAI backer Microsoft, Alphabet recently saw the division’s sales increase.

Starting Thursday, Google announced it would open up its million-token AI to a limited number of enterprise customers, while any developer could build with version 1.0 and switch to the latest generation when it’s available.

Asked when such powerful, typically cost-intensive AI models would be profitable for Alphabet, Pichai said: “These are profitable things for us. Over time, we will also be very efficient in using these models.”

Efficiency was another improvement Google highlighted in Gemini 1.5. Google said it introduced a less burdensome way for the AI to gather information, known as “expert blending,” in line with its previous research and a recent announcement by a rival.

One company official likened the approach to calling a fortune teller to answer a question, rather than calling every person you know.