I agree there's no single definition, but I think they all have something current LLM don't: the ability to learn new things, in a persistent way, with few shots.

I would argue that learning is The definition of AGI, since everything else comes naturally from that.

The current architectures can't learn without retraining, fine tuning is at the expense of general knowledge, and keeping things in context is detrimental to general performance. Once you have few shot learning, I think it's more of a "give it agency so it can explore" type problem.