Having it written in smaller, harder to read font and positioning in the place that would typically contain photographers name which everyone usually ignores is opposite of highlighting.

If it was used to replace generic stock image with beer, I might be fine with it. As an illustration generic picture of beer cup without powder or branded packaging would have been sufficient. But being so specific it contains fake brand, powder makes it seem like it represents the actual product thus giving a misleading impression about the structure of powder and it's capabilities to produce foam.

Why do you care, or even know that it “brings down the cost of the article”?

> I thought it helped to visualise the product, brings down the cost of the article.

Does it though? First thing I thought when I opened the page and saw that image was that the granules looked quote large and that it would take some effort to make them dissolve nicely. Why have an photo realistic image if its not an actual photo, that might or might not missrepresent the product.

Why not take the bring the cost down to its logical conclusion and have AI write the story from the press release and other content on the internet?

Pictures in journalism are information, just organized differently than text. If you use a random generated image that “mimics” that information you are effectively doing misinformation or even disinformation. No picture (or a generic picture of the monastery) would have been better in terms of information content. This publication decided to go for a subpar solution because it’s what? Cool to some?. Even worse: actual images were available.

I made very clear I am not suggesting sending a human out to do the job, so the cost argument here doesn't fly. As I said already I would be satisfied if the author used a simple google image search or wikimedia search.