machine learning theory

Is AGI development hitting an unexpected wall with 'common sense' reasoning?

I've been following AGI progress closely, and while models excel at complex tasks, they often stumble on basic, intuitive human common sense. Are we fundamentally missing something...