Microsoft says new ChatGPT shows signs of being artificial intelligence with "human capabilities"

247 - Um article A study published last week by Microsoft researchers, titled "Sparks of General Artificial Intelligence: Early Experiments with GPT-4," raises the possibility that... the new "brain" model of ChatGPT, the GPT-4It could be considered a primitive form of Artificial General Intelligence (AGI) – that is, the model exhibits superhuman abilities in various areas of knowledge. information The story about the article was reported by journalist Bruno Romani, from Estadão.

Currently, Artificial Intelligence systems are categorized as "narrow AI," meaning they excel at very specific tasks. On the other hand, an AGI is a class of systems with exceptional performance across multiple tasks, something closer to what science fiction often portrays. Microsoft researchers conceptualize AGI as a system capable of reasoning, planning, problem-solving, abstract thinking, understanding complex ideas, and learning from experience.

>>> Five tasks that GPT-4 can perform that ChatGPT couldn't.

"Given the breadth and depth of GPT-4's capabilities, we believe it could reasonably be viewed as an early (albeit incomplete) version of an AGI system," the researchers stated.

The researchers describe in their article that they proposed a series of tests for the GPT-4, and it demonstrated an 80% success rate on the US medical licensing exam and 70% on the US version of the "OAB exam" (Brazilian Bar Exam). The system also had 100% efficiency on the test administered by Amazon to software developers, completing the exam in just four minutes.

"We have demonstrated that, in addition to mastering the language, GPT-4 can solve original and difficult tasks, including mathematics, code creation, vision, medicine, law, psychology, and more, without the need for special commands," the article states.