Google Presents A Fast And Efficient New Artificial Intelligence Model

JAKARTA - Google has introduced Gemini 1.5, an upgrade to a series of artificial intelligence (AI) models announced two months ago. In one request, users can request Google AI to analyze much more content than previously - 30,000 code lines, 11 hours of audio, or a full hour of video.

The Pro model, the family's medium size, can process 1 million data tokens - five times the amount that Anthropic can handle, perhaps the closest competitor in this regard, with Claude 2.1 technology.

Google's increasingly greedy AI, along with releases from OpenAI and others, reflects the fast-going arms race in Silicon Valley to build the most capable technology - and most marketable.

In an interview with Reuters, Alphabet CEO Sundar Pichai welcomed this progress as one of the "breakups" that would support his company's various business. Basically, he said, the founder of the search industry made a new type of investigation possible. "You are dramatically giving wider views to people to ask about the world," Pichai said.

According to Pichai, one of the examples Google is talking about internally is how a filmmaker can ask AI to assess abusive films like a critic would. "This is just one example of the use we talk about as a team, but the sky is the limit here," he said.

Another: submitting several company financial reports at once, Pichai said. Press demonstrations show how AI can extract information from 44 minutes of video in about 59 seconds, as well as multimodal requests, in which a user asks AI to respond to a combination of text and images.

In addition to providing updated technology to Google Search and YouTube, Alphabet also hopes Gemini 1.5 will attract customers to its cloud units. In competition with larger rivals like Microsoft, Alphabet recently saw an increase in sales growth for the division.

Starting Thursday, February 15, Google said it would open the millions of token-based AI for a limited number of business customers, while each developer could build with version 1.0 and switch to the latest generation once available.

Asked when such a powerful AI model would benefit Alphabet, Pichai said: "These are beneficial things for us to do. Also, over time, we will be very efficient in running these models," he said.

Efficiency is one of the other improvements Google highlighted for Gemini 1.5. Google says implementing less burdensome ways for AI to gather information known as "interventional experts," according to previous research and a recent announcement of competitors. A company official likened the approach to calling an irony to answer questions, rather than calling everyone we know.