Not necessarily, technology development isn't linear nor steady. Right now we are at peak "gold rush" stage of the cycle. Investor money is subsidizing the development and the operation of the AI models. Outside of very specific applications, no real "killer app" use has materialized to justify how expensive it is to run the larger models. Right now they are the cheapest they'll be to the general user. Open Ai loses money with every request made by paying users. And we are getting to the point where stakeholders are starting to demand RoI on all the AI investments over the last few years, so investor money is bound to fry up soon. So far, only Nvidia is making bank while everybody else just keeps spending hand over fist.
And then there's the physical and data limits of the technology. Chips are getting closer and closer to the smallest size they can be, while the energy requirements to keep the data centers running increase year after year straining electrical grids. And then they are running out of clean data to train new models. (Just the smallest amount of data not made by humans can collapse a model hard) The current approach to AI is a technological dead end that will leave some useful stuff in the process, but not the road to Skynet many fear (or hope). Outside of the most trivial tasks -or some of the most specific and domain restricted models-, models output has to be cleaned up and processed by a human to be useful, to the point it is a full-time job in many places.