The market isn't growing due to demand, though. AI is being pushed by businesses and investors hoping that demand will appear in the future. They're burning through tons of cash, not growing naturally due to user interest.

While it's true that some demand is related, the vast vast majority of the spend is speculative on what AI might be and assuming that if it manages to be the thing of their dreams, it will demand exponentially more resources than the current LLM slop.

They are spending this money without clear indication that the demand they want is there. For example, OpenAI at one point claimed that, by now, businesses would be paying them $50k/year for a single 'instance' of LLM, good for equivalent to one human headcount.

They are currently betting that at some point, they'll effectively fix the lack of actual reasoning (a number of AI enthusiasts will claim that AI can have an entirely distinct thing from any reasoning we have ever known but still call it reasoning, which is a pretty stupid cop out). And/or they'll translate this reliably to robotics (so far this has ended up being pretty elusive, investors assumed the same fake language that passes for executive-speak means it could apply to menial manual labor, but it hasn't worked yet).

But yeah, refraining from using these services would deflate the expectations more quickly, enough people toying around with it sparks the imagination of what the execs think they can get for it...