In a product update that is closely watched as a gauge of whether generative AI is progressing quickly or plateauing, OpenAI unveiled the fifth iteration of the AI engine that powers ChatGPT on Thursday.
GPT-5 is released more than two years after GPT-4 was released in March 2023, marking a time of significant business investment, excitement, and concern about the potential of AI.
As part of a battle with Google and other rivals in the U.S. and China to outperform one another in AI standards, rival Anthropic earlier this week unveiled the most recent iteration of its chatbot, Claude, in anticipation.
Microsoft, a longtime partner of OpenAI, announced that it will integrate GPT-5 into Copilot, its AI-powered assistant.
The San Francisco-based business has long positioned its technological achievements as a path toward artificial general intelligence, or AGI. This technology is expected to outperform humans at economically meaningful tasks, so expectations are high for the latest iteration of OpenAI’s flagship model.
To achieve this, it is also attempting to raise a significant amount of money, partly to cover the expensive data centers and computer chips required to develop and operate the system.
The new model is a “significant step along our path to AGI,” according to OpenAI CEO Sam Altman, but he mostly focused on how easy it is for the 700 million people who use ChatGPT every week.
At a debut event that was livestreamed on Thursday, Altman remarked, “It’s like talking to an expert—a legitimate PhD-level expert in anything, any area you need, on demand.”
The new model, which is currently accessible to anyone with a free ChatGPT account, with usage limits, may take some time to observe how people utilize.
The CEO of coding software manufacturer Cursor, a significant Anthropic client, made a special appearance at the Thursday event, which centered largely on ChatGPT’s application in coding, an area in which Anthropic is regarded as a leader.
The presenters from OpenAI also discussed safety enhancements that would make the chatbot “less deceptive” and prevent it from generating dangerous answers to “cleverly worded” requests that would get past its safeguards.
According to a recent study, ChatGPT was giving researchers who were pretending to be teens harmful material on drugs and self-harm.
Technically speaking, GPT-5 demonstrates “modest but significant improvements” on the most recent benchmarks, but it also looks very different from GPT-4 and resets OpenAI’s flagship technology in a way that may lead to future advancements, according to Cornell University assistant professor of computer science John Thickstun.
He stated, “I don’t think AI will simply solve all of humanity’s problems and that it’s the end of work, but I do think there’s still a lot of headroom for them, and other people in this space, to continue to improve the technology.”
“Making the most of the gains already made is not enough.”
Originally founded in 2015 as a nonprofit research lab to develop AGI responsibly, OpenAI has subsequently grown to become a for-profit business with a $300 billion valuation.
Since Altman’s removal from the nonprofit board in November 2023, the corporation has attempted to alter its organizational structure. After a few days, he was reinstated.
The California and Delaware attorneys general, who have jurisdiction over nonprofits, have been investigating the matter, and Elon Musk, the founder and early donor to OpenAI who now owns an AI company, has filed a lawsuit against it. Despite this, it has not yet disclosed any profits.
Most recently, OpenAI announced that it would convert its for-profit business into a public benefit corporation, which will have to strike a balance between its purpose and shareholder interests.
According to banking behemoth JPMorgan Chase, which recently made the unusual decision to cover the firm despite it not being publicly traded, OpenAI is the third most valuable private company in the world and a bellwether for the AI industry, with an “increasingly fragile moat” at the frontier of AI.
The bank stated in its report last month that businesses may be forced to compete by reducing the pricing of their AI products as a result of the impossibility of a single AI developer maintaining a “sustained competitive edge.”
