We've had AI booms before. In terms of capabilities, this one is following exactly the same trajectory. Human researchers come up with some sort of breakthrough

We've had AI booms before. In terms of capabilities, this one is following exactly the same trajectory. Human researchers come up with some sort of breakthrough improvement to AI methods, that results in exponential like growth of capability as we both solve the low hanging fruit the method offers at it, and scale up the compute and data available to the limits of what is useful for the method. Then, capabilities start to plateau, and there is a long tail of the new techniques being applied in specific situations as they get combined with domain specific tuning and architecture. We are well into this process already. Core chat capabilities have pretty much stalled out. But most of the attempts at application are still very thin layers over chat bots.