The startup OpenAI on Tuesday said it has started to launch a new artificial intelligence model called GPT-4, setting the stage for the human-like technology to become more and more competitive between its backer Microsoft Corp and Alphabet Inc’s Google.
OpenAI, which created the chatbot sensation ChatGPT, said in a blog post his latest technology is “multimodal,” which means that images as well as written information can inspire creation. The installation section will be available in ChatGPT Plus and programmers, with a waiting list, while still using images as a display of his studies.
The launch is expected to show how office workers can turn to AI improvements for many jobs, as well as how technology companies will lock in the competition to win business. from progress.
Alphabet Inc’s Google on Tuesday hearing a “magic wand” for its collaborative software that can write almost any document, days before Microsoft is expected to introduce AI for its competitive Word processor, possibly powered by OpenAI. A Microsoft executive also said that GPT-4 is helping to power its Bing search engine.
The new technology of OpenAI in some cases showed an improvement over a previous version called GPT-3.5, it was said. In an example of the skin test required of American law school graduates before taking up professional work, the new model scored in the top 10% of tests, compared to the old model in the top 10% bottom line, says OpenAI.
The OpenAI logo is seen in this photo taken February 3, 2023. REUTERS/Dado Ruvic/Mirror
While the two versions may look similar in general conversation, “the difference comes when it comes to the complexity of the task at a reasonable level,” OpenAI said, noting that “GPT-4 is more reliable, creative, and able to handle a large number of numerical orders. .”
An online demonstration of the technology by Greg Brockman, the president of OpenAI, showed that it is possible to take a picture of a hand model for a simple website and create a real website based on it. GPT-4 can also help individuals to plan their taxes, the presentation revealed.
Sam Altman, the CEO of OpenAI, on Twitter called GPT-4 his model “the most capable and compatible” with the values and expectations of people, although “it is still flawed. “
GPT-4 has 82% less response to unauthorized content requests than its predecessor and scores 40% higher in some reality tests, the company said. Incorrect responses known as “fake” have become a challenge for many AI programs.
Microsoft stands to benefit from the adoption of GPT-4, said Rishi Jaluria, an analyst at RBC Capital Markets.
The developer is not only integrating OpenAI’s new technology into its products: its Azure cloud is encouraging the use of OpenAI as well as businesses. budgets are looking at IT spending in an uncertain economy, he said.
“Whenever a company uses this piece of technology,” said Jaluria, “those jobs go to Microsoft Azure, and I think it’s going to come very soon. difficult.”