How is AGI different from current AI?

Current narrow AI systems are much more domain-specific than AGI. We don’t know what the first AGI will look like. Some people think that scaling up the GPT-3 architecture may be sufficient to produce AGI, while others disagree. (GPT-3 is a giant prediction model which, when trained on a vast amount of text, seems to learn how to learn and do all sorts of highly-impressive things; a related model can generate pictures from text.)