Microsoft researchers recently published a research paper on GPT-4, claiming that it exhibits;"sparks" of artificial general intelligence (AGI);or cognitive abilities on par with humans.
;GPT-4 performs tasks remarkably close to human-level performance in various domains such as mathematics, coding, vision, medicine, law, and psychology, according to the researchers.
- The;paper, titled;"Sparks of Artificial General Intelligence: Early experiments with GPT-4," highlights GPT-4's potential as an early version of an AGI system while acknowledging the need to identify its limitations and overcome challenges in advancing towards more comprehensive AGI systems.
- The researchers were particularly impressed with GPT-4's solution to a;reasoning exercise involving stacking objects.
- They asked both AI models to stack a book, nine eggs, a laptop, a bottle, and a nail in a stable way. While GPT-3 seemed confused, GPT-4 surprised the researchers with its ingenious approach.
- GPT-4 suggested arranging the eggs in a three-by-three grid on top of the book, creating a stable platform for the laptop and other objects to balance.
- It explained that the laptop would securely fit within the book and eggs, providing a stable surface for the next layer.
- "All of the things I thought it wouldn't be able to do? It was certainly able to do many of them — if not most of them," lead author Sébastien Bubeck told The New York Times.