Microsoft is clear about it and in a latest report it talks about its AI, GPT-4, being an early form of artificial general intelligence (AGI).

Following the release of the new OpenAI language model, GPT-4, which will enhance the skills of ChatGPTa team of scientists from artificial intelligence Microsoft has published a research paper claiming that the model, which already powers Bing’s AI, displays traits of human-level intelligence, or artificial general intelligence (AGI).

The article titled “Sparks of Artificial General Intelligence: First Experiments with GPT-4”, reports that GPT-4 has shown early signs of AGI, meaning that has capabilities that are at or above the human level.

The researchers argue that GPT-4 is stronger than other OpenAI models that have preceded it in new and widespread ways.

“We show that, beyond its language proficiency, GPT-4 can solve novel and challenging tasks spanning math, coding, vision, medicine, law, psychology, and more, without requiring any special prompting.”the researchers write in the abstract of the article.

Microsoft claims GPT-4 shows ‘sparks’ of artificial general intelligence (AGI): ‘clickbait’?

“GPT-4’s performance is surprisingly close to human-level performance and often far exceeds previous models like ChatGPT. Given the breadth and depth of GPT-4’s capabilities, we think it could reasonably be seen as a early (but still incomplete) development of an artificial general intelligence (AGI) system”they explain.

They conclude that this artificial intelligence overcame some basic problems, such as acquiring many non-linguistic abilities, while making great progress in common sense.

“Our claim that GPT-4 represents progress towards AGI does not mean that it is perfect at what it does, or that it comes close to being able to do anything a human can do (which is one of the usual definitions of AGI) , or who has internal motivation and goals (another key aspect in some AGI definitions)”reads the document.

It is true that most of the document contradicts itself since it is dedicated to listing the number of limitations and biases that the language model contains, so it begs the question of how close it really is GPT-4 of these capabilities and how perhaps the AGI concept is being used instead as a clickbait.

Keep in mind that Microsoft researchers may have a vested interest in promoting OpenAI’s work, as the company partnered with it for billions of dollars earlier this year, so it would not be surprising if they were hyping up this new model which will power ChatGPT.