For a new Google artificial intelligence model, reading “War and Peace,” Leo Tolstoy’s massive work that is half a million words lengthy and frequently more than 1,000 pages long, might take only a minute or two.
The business, owned by Alphabet, unveiled Gemini 1.5 on Thursday, an update to its AI model suite that it had disclosed two months prior. Users may now ask Google’s AI to evaluate far more stuff than they could previously with just one prompt: 30,000 lines of code, 11 hours of audio, or an entire hour of video, according to Google.
That pertains to the mid-sized Pro model in the series. To put it another way, in a query of its Claude 2.1 technology, Pro can analyze one million tokens, or data points, which is five times more than Anthropic, which is probably the closest rival in this area.
Reuters was unable to verify the accusations on its own.
Following releases by OpenAI, the company that created ChatGPT, and other companies, Google’s consuming AI is a reflection of the intense competition in Silicon Valley to develop the most advanced, commercially viable technology to date.
Alphabet CEO Sundar Pichai hailed the development as one of several “breakthroughs” that would power his company’s diverse industries in a Reuters interview. Fundamentally, he stated, the mainstay of the search industry was enabling a new mode of inquiry.
“You’re dramatically giving a wider view for people to ask questions about the world.”
According to Pichai, one idea that Google addressed internally was how a filmmaker could ask the AI to evaluate a rough-cut film in the same way that a reviewer would. “This is just one use case we were talking about as a team, but the sky’s the limit here,” he stated.
Pichai mentioned another: simultaneously requesting financial reports from multiple companies. Press demonstrations demonstrated how the AI could perform multimodal prompts, where a user asked the AI to respond to a combination of text and pictures, and how it could extract information from a 44-minute movie in around 59 seconds.
In addition to providing YouTube and Google Search with new technology, Alphabet hopes that Gemini 1.5 will attract users to its cloud division. Amidst intense rivalry with Microsoft, which supports OpenAI, Alphabet has witnessed a resurgence in its division’s sales growth.
Google announced on Thursday that it will make its million-token AI available to a select group of business clients. Developers may continue to use version 1.0 of the AI and upgrade to the most recent version as soon as it became available.
When asked when Alphabet will see a profit from such potent AI models, which are usually expensive, Pichai replied, “These are profitable things for us to do.” We will also become extremely proficient at using these models over time.”
Another enhancement Google emphasized for Gemini 1.5 was efficiency. Google added that, in keeping with its previous research and a competitor’s recent disclosure, it has introduced a less burdensome method for the AI to obtain information known as a “mixture of experts.”
An executive from the company compared the method to asking a genius to respond to a query instead of phoning everyone they know.
