AI is probably going to be revolutionary, but the question is what kind of a revolution it's going to be. It all depends on how it's used, and that's something that's hard to foresee yet.
I'm not particularly convinced of LLMs, because they're actually really dumb. They can produce some impressive results, but in the end, they're unreliable, which severely limits their usefulness. They're neat and useful, but they have their limits. Perhaps they can be augmented to eliminate those weaknesses, but if I had to guess, I'd say it's a huge undertaking.
In the end, my current bet is that AI will not revolutionize anything any time soon, despite some impressive advances in the recent years. Currently I'm thinking it's going to be gradual advancements across different fields instead of a sudden leap revolutionizing everything at once, so there might not even be a single point of revolution. Instead, it might be something that when you look 30 years back at some point in the future, you see how vastly everything has changed, but it's across those 30 years that it's happened instead of very quickly (not that 30 years isn't very quick, relatively speaking, but you know what I mean - and no, 30 years isn't a proper prediction; it's more like a first guess).