> The problem comes when the imagery is placed as a false value signal. If it is designed to send a signal of "this content took a lot of effort to produce because an image that would be hard for a human to make is attached to it" - but the image was generated without that effort, then it's basically a lie and insult to the reader.

Well, the issue is the majority of AI images seems to fall into this category, which then trains my brain to treat all AI images in this way, rather than sifting for the few golden use cases where it's legitimately adding value.

At work, for example, I'm more likely to go brains off if someone presenting content shows an AI generated image in their slides.