A lot of components that, potentially, are going to have to enter for AGI to be achieved.

Glorified search is how I look at these chatbots.  They are a major improvement over search engines but that is all.  Does that mean we will not get the next generation making even greater strides?  I wouldn't bet against it.

However, to me, there needs to be a couple more things for AGI.  One is spatial since it is not likely to be able to form any general sensibilities without a clear understanding of space.  Second, I think some form of symbiotic learning is required.  It is one thing to guess the token probabilities.  However, unless it can comprehend the similarities (and differences) across many different disciplines, we are just mashing together more data and faster compute.

This could take us far especially as compute increases but will not, in my view, come close to AGI.