It's possible, sure ...
Is it a cool, ethical, or respectable decision? No, not really. People are free to use AI if they want, but they are not owed the unconditional respect of the creative community (which skews anti-genAI, for very good reason, pretty heavily) in doing so, and potential customers are similarly free to reject them and even criticize them for doing so. The product will be seen as lesser and, in my opinion, this will be because it is, in fact, lesser, both in quality and in terms of more abstract intrinsic value. But, my biased opinion aside, it's likely going to negatively impact your sales potential, which, if you're working in the indie space, is already more or less in the gutter (so, I guess it's not that big a deal - if you're not planning to make any money anyway, then there's really no fiduciary pressure, though there may still be some reputational drawbacks).
If you can't afford human artists for bespoke commissions, there are a lot of human made creative commons artwork out in the world, made generously available by the artists who made it with their own hands specifically for this reason, not to mention the absolute plethora of public domain artwork (I suspect that fantasy games in particular should have a veritable treasure trove of early pulp illustrations to draw from already, and that trove is going to grow exponentially in the coming years) out in the world. In my observation, players are going to be much more open to and forgiving of public domain solutions when compared to AI slop, so it's the better (free) option from a purely business perspective, in my opinion, even if you are pro-AI, or have no strong opinions on the matter one way or the other. With good curation / art direction, as well as some solid graphic design chops, one can use Creative Commons and Public Domain artwork to great effect and end up with beautiful and unique transformative products are a result.
Personally, assuming your question is related to potential reception of the product, as opposed to mere classification (I did find it a bit unclear, so I'm sorry if I've misconstrued overall), I don't recommend doing it. Ethical and environmental matters aside, I don't think large scale AI, of a kind capable of generating imagery, is ever going to be "cool" in the way that the CEOs hope and wish. It appears that proponents are waiting for a watershed moment, or a critical mass of adoption, on which they can then market the product to its actual intended clients for its actual intended purpose (the former being governments and government contracters, and the latter being mass data gathering, surveillance, and control), but it's just not coming fast enough ... mostly because those same proponents are themselves profoundly, deeply, unpleasantly uncool people (see Musk, Thiel, Zuckerberg ... hell, even Trump, who has been personally slinging slop in a public and unflattering way).
It might be "here to stay" (even that, I am not too sure of, given the bizarre and precarious economic situation, but let's assume it somehow manages to survive intact without the bubble popping), the grass roots adoption is primarily skewed towards boomers (who are not cool) and conservatives (not cool). The "high level" adoption and marketing is primarily associated with rich, white, conservative dudes (very, very uncool). In other words - facebook is probably "here to stay" too, but we all know who its for, who the current primary users are, and while some of them are our parents and grandparents ... well ... let's face it, it's pretty much all parents and grandparents. In other words, even outside of any moral considerations, AI has a distinctly boomer face to it, and that's not likely to change as the new generation comes through (when was the last time kids were falling over themselves to get on board with the stuff their grand parents are into).