OpenAI is ending October with a new profit structure, a new deal with Microsoft, and a whole new level of pressure to achieve artificial general intelligence.
The generally accepted definition of AGI is an artificial intelligence system that equals or exceeds human cognitive ability. But even as it has become a goalpost for the entire AI industry, people disagree on what it actually looks like and if or when it might become possible. A large part of the economy is devoted to finding the entirely imaginary finish line of this expensive race. And after Tuesday’s news, there’s a new entrant in the race: Microsoft itself.
In 2019, OpenAI and Microsoft entered into a partnership governed by an ambiguous “AGI clause”. The deal announced that Microsoft had the rights to use OpenAI’s technology, but only until OpenAI achieved AGI… whatever that means. OpenAI had a clear incentive to announce that it had acquired AGI as quickly as possible in order to stifle competition and gain control over its own IP. Microsoft had a similarly clear incentive to add friction to that process, even though it had supported AGI as the overall goal of AI.
OpenAI’s restructuring depended on Microsoft’s green light – and OpenAI would have lost. as much as 10 billion dollars If the conversion to profit did not occur by New Year’s Eve. Microsoft appeared to be getting several concessions in the resulting negotiations. As part of companies new Deal“Once an AGI is declared by OpenAI, that declaration will now be verified by an independent expert panel” – This will not be OpenAI’s decision alone. And now, Microsoft doesn’t even lose all its rights on the technology. According to the new deal, “Microsoft’s IP rights for both models and products have been extended to 2032 and now include subsequent AGI models with appropriate security guardrails.”
However, one of the most significant changes was that Microsoft can now “independently pursue AGI alone or in partnership with third parties” – meaning the AGI arms race continues in a way it never has before. Things are about to get more intense than ever, as absurd as it may sound.
For one thing, Microsoft is perfectly within its legal rights to use OpenAI’s IP to develop its own AGI and try to win the race. Yes, it will face some consequences: if it uses OpenAI’s IP to achieve the milestone, the models will be subject to certain computation limitations until AGI is officially announced. But remember, Microsoft has full IP rights to OpenAI’s research (at least, until 2030 or when the expert panel verifies that AGI has been reached, “whichever comes first”). This means Microsoft can only access OpenAI for internal use, or for research purposes, as well as “confidential methods used in the development of models and systems.”
Microsoft may also work with one or several of its OpenAI competitors in the race to take on AGI. Last month, Microsoft Started buying AI From OpenAI’s arguably biggest rival, Anthropic, to diversify its AI partnerships; We may see the company working with Anthropic or any other startup in the near future.
The risks of declaring that humanity has achieved AGI are enormous. Plus, the deal is coming after months of AI developers shifting the goalposts, often in ways that suggest AGI is just around the corner. OpenAI CEO Sam Altman has hinted at AGI coming by 2025 or at least in the next few years, Write In January, “We now believe we know how to create AGI as we have traditionally understood it.” Anthropic CEO Dario Amodei has predicted He says “powerful AI,” a term he prefers to use in comparison to AGI, “could happen as early as 2026, although there are ways it could take longer.” It’s a term we’ve heard from executives on earnings calls, a question we’ve seen on message boards for tech CEOs, and in countless memes.
One thing that OpenAI gained from the new Microsoft deal, besides the ability to move forward with restructuring to its advantage, is the fact that Microsoft does not have access to IP for its consumer hardware. That means the startup is likely betting big on its upcoming AI device, which it is working to develop with famed former Apple designer Jony Ive. Altman emphasized this during Tuesday’s livestream, highlighting that OpenAI wants to create “a personalized AGI that you can use anywhere, to help you at work and in your personal life.” The new goal that OpenAI highlighted was part of a growing trend by companies to limit their focus to “personal AGI” and consumer hardware, similar to Meta CEO Mark Zuckerberg’s own AGI manifesto, which promotes “personal superintelligence.”
The idea of a panel determining whether OpenAI has achieved AGI makes sense in many ways. AGI is such a buzzword that it is not necessary for a single company to announce that it has reached a milestone without any outside input, especially when such an announcement would have huge implications, especially in the financial markets. Even Altman said during Tuesday’s livestream that the term AGI has become “extremely overloaded” and no longer means as much as it used to. It is therefore appropriate that a panel of experts should help define it. Of course, the question will be who selects those experts. OpenAI and Microsoft have not released any details about how the panel will be selected, and OpenAI declined to provide any further details when contacted. The Verge,
Altman himself said in a Q&A with reporters at the company’s annual DaveDay earlier this month that “there are many parts of AI that I think are kind of bubble-y right now.”
During Tuesday’s livestream, chief scientist Jakub Pachocki said that as companies enhance deep learning and develop new algorithms, AI will eventually surpass some aspects of human intelligence on the intelligence curve and “has already overtaken it on several axes.” But, he said, achieving it does not mean that you have solved all the problems around it, which we need to think seriously about.
“This (AGI) process will take many years, which we’re in the middle of,” Altman said during the livestream. “I think it’s more useful to say that our intention, our goal, is to have a true automated AI researcher by March of 2028 and to define what that is to try to satisfy everyone with a definition of AGI.” But there’s at least one party that OpenAI would actually have to be satisfied with that kind of definition: Microsoft. And billions of dollars are at stake.
