It appears that Microsoft and OpenAI have defined artificial general intelligence, or AGI, in a very novel way as part of their agreement to let Microsoft use OpenAI's AI models. Under their definition, AGI is achieved when OpenAI can turn a profit of $100 billion from its AI systems.
The definition is far from what anyone would consider a good definition of AGI. Typically, people would define AGI as an AI that can do a broad range of tasks as humans can. There are no fixed definitions of what AGI is, so there are disputes about whether it has been achieved or not; some people even believe we are at AGI, while others say it's decades away.
The definition that Microsoft and OpenAI arrived at is probably so that they can have a measurable milestone. According to TechCrunch, having a measurable AGI definition is very important because, at this point, Microsoft will lose access to OpenAI's models and must have developed its own by then to compete.
To be clear, Microsoft is not waiting until this milestone is reached to roll out its models. Just a few days ago, Neowin reported that the Redmond company is going to deploy in-house large language models (LLMs) for its Microsoft 365 Copilot to reduce its reliance on OpenAI.
Microsoft's and OpenAI's definition is certainly interesting and will likely be added to a roster of AGI definitions to track humanity's progress towards that objective. If OpenAI ever decides to go public, its profits will be available for all to see on a quarterly basis so we will have a good idea of how close to AGI we are, according to this metric at least.
Let us know in the comments how you define AGI, do you like Microsoft's and OpenAI's definition?
Source: TechCrunch via The Information
_small.jpg)
4 Comments - Add comment