This is partly true, but the prediction isn't based on the previous token. It is based on the tokens generated, in the context of having parsed trillions of tok

This is partly true, but the prediction isn't based on the previous token. It is based on the tokens generated, in the context of having parsed trillions of tokens and calculated the relative importance of those tokens in the context of the tokens around it. So the matrix of weights not only encodes probability of the next work, but also encodes which word (amongst a choice of equally probable words) makes the most sense in the token stream. That's my very shallow understanding of the attention mechanism anyway.