With ChatGPT, models for artificial intelligence (AI) have found their way into various areas of life – but they do not fulfill what their inventors understand by the term general artificial intelligence, including in financial terms. Confidential documents now provide an insight into an agreement between the (still) non-profit organization and the investor and business partner Microsoft. According to the agreement, both companies consider an AI model from OpenAI to be “Artificial General Intelligence” (AGI) – as soon as it breaks the 100 billion dollar profit mark.
Anzeige
AGI is a theoretical, particularly powerful form of artificial intelligence. Microsoft and OpenAI – provided two definitions on their own websites even before the current revelations –: OpenAI speaks here of “highly autonomous systems that outperform humans in most economically important tasks.” According to Microsoft, AGI would be “capable of performing a variety of tasks at or above human level” and “the next logical step after current AI technologies such as Copilot, which can already perform a wide range of tasks such as text creation, image generation, data analysis and much more.” There are many different opinions on the exact point at which a technology is considered to have fulfilled these requirements – as well as on how much AGI use the future could really be characterized by.
AGI definition is controversial for OpenAI
Microsoft’s and OpenAI’s views on this are probably primarily of a financial nature, as research by the news portal The Information shows. According to this, the two companies agreed as early as 2023 that they would consider an AI model from OpenAI to be an AGI model as soon as it generates at least 100 billion dollars in profit. Also explosive: according to the contract with OpenAI, Microsoft will lose access to the start-up’s technologies as soon as it reaches the status of General Artificial Intelligence, as the New York Times reported in October, citing 19 anonymous sources at OpenAI. The clause is intended to prevent Microsoft from misusing the AGI technology.
However, OpenAI is a long way from a 100-billion-dollar profit. The tech start-up is expected to make a loss of around five billion dollars in 2024. In a financing round, the tech startup, which still operates as a charity organization, received fresh money from investors such as Microsoft, Nvidia & Co. However, this is linked to the condition that OpenAI becomes a for-profit organization. On Friday, OpenAI announced its intention to transform itself into a Public Benefit Corporation (PBC). Microsoft and OpenAI are also business partners: Microsoft provides the AI developer with the necessary computing power, while ChatGPT is a component of Microsoft’s own AI application Copilot. There have recently been media reports that Microsoft is also trying to use other AI models from other companies for the application.
If it really takes a 100 billion dollar profit to release OpenAI from Microsoft’s grip, it could be quite a while before OpenAI comes up with a real AGI model. On the other hand, in the medium or long term, Microsoft could equip its AI applications with something that is equivalent to an AGI model, but which those responsible are not yet allowed to call it that – for financial reasons.
(nen)
