It’s like all new tech, if you fall into the survivorship bias trap and only look at the ones that have survived.

Personally I think this will be another “AI winter”, where we will gain some improvements in narrow domains but fall far short of what was promised, leading to disillusionment and a reduction in research budgets.