Literally did that yesterday with ChatGPT. Used it as tech support for changing my display refresh rate via a shortcut command. It kept apologising and offering different solutions, not losing its temper once. 10/10 would use it over a real person.

It's true, I've also used the likes of Unsplash, but with AI art, the ownership falls onto the person that gave the prompts in the first place, potentially shifting the value from "cheap" to basically "free" in the market. Why pay the artist, when you can describe what you want, get it done in mere minutes and only pay the price of electricity your PC uses, while having full ownership of the generated artwork? As AI evolves to be more creative, detailed (and also better in drawing text), I believe this will be a real problem over time, starting from generic work, as "placeholders".

I think that for more specific and extensive work, artists won't be out of a job any time soon. But for smaller projects, like making a personal webpage, will benefit greatly from AI art, as being generic won't be a problem. (I'm also speaking from my own experience, where I was able to generate a decent image for my architecture project in Stable Diffusion within minutes, instead of Googling the image for hours)