I think that in many cases, people want their own tastes affirmed. That’s not a knock; it’s just human nature. People are the same way with other sorts of video games, films, comic books, etc. In many cases, a review simply can’t please everyone. I don’t think it should, either. Or that shouldn’t be a goal, since reviews are mostly opinion (even when we’d like them to be objective). I often value minority positions more than I do conventional wisdom, so long as they’re well argued.
Somebody didn’t like one of my Zork reviews. I said some things that could be called controversial. That’s ok. I gave the subject a lot of thought, and I would say the same things today. If people disagree, I don’t mind.
Thanks for the replies. I’ve realized that I shouldn’t focus on it too much, and there’ll always be people who disagree with me or my comments. It’s not worth worrying about, and a lot of my reviews get more positive responses than negative, so I want to keep writing them.
And yeah, comments would be a nice feature. Noticing the controversial aspect of the game kind of deterred me from playing any more, since it wasn’t in the description anywhere and I wanted potential players to be aware. The best way to do that just seemed to be an unstarred review, and the IFDB guidelines even mention you can do this if you haven’t gotten far enough into the game to form a full rating.
If I’m writing a 3 star or lower review I brace for the possibility of one unhelpful vote since not all authors take criticism well. If most of my reviews are mostly considered helpful then not then that’s what’s important to me.
Looking at the review that I think you’re talking about, it’s easy to infer your opinion on AI art from it (and I believe it’s one I share). Since it’s a controversial topic the “unhelpful” votes could possibly come from people who disagree with your opinion rather than people who dislike the review. Since they’re anonymous they’re easily abused.
I don’t have empirical data on Goodreads, but the similar site LibraryThing provides very extensive stats on the books you log and also has a “site vs. user” tab where you can see how your stats compare to the site as a whole. This is my site (grey bars) vs. user (colored bars) graph for star ratings:
Leaving aside whatever this says about me (I would’ve showed just the site stats if LT had that option, but I don’t think it does), you can see that on LibraryThing, at least, the 3-, 4-, and 5-star ratings are used far more than the 1- and 2-star. The two lowest full-star ratings*, combined, make up less than 10% of the total number of ratings on the site. It’s not necessarily generalizable to Goodreads, of course, but I have also anecdotally observed that most books on GR have an average of somewhere between 3 stars and 4.5 stars, which is consistent with the distribution of ratings seen on LT.
** The half-star ratings are super wonky both for me and for LT in general because I, like a lot of people, got started on LT by importing my GR reading history, and GR doesn’t do half-stars.
Edit: Of course, I realize that this is completely off-topic, I just got excited about statistics.
I think I have tons of reviews in this territory (most of them?). Love it but technically it’s full of holes… Technically it’s great but I didn’t dig it… etc. But I can explain any combinations like these in my writing. Then I hope the stars plus the writing convey something together on their own. And then together en masse in my output. And that in the aggregate, they feed the overall picture.
If you’re concerned about correctly expressing yourself with these kinds of ratings, I figure writing a review is the way. If you’re unconcerned, just trust that your numbers help the big picture, though what they mean just on their own (what did bg make of this game?) might be fuzzier.
I’ve rated 76 games on IFDB (mostly old ones), but I’ve never written a review there. If reviews were required for ratings, I might not have any ratings. I guess it depends on what people want from IFDB.
I’m a player, not an author, so I don’t easily relate to authors’ concerns. I have emailed feedback to IFComp authors before, but it’s kind of none of an author’s business what IFDB says about their game. I guess if you’re the kind to make a bug-fix release (I don’t see the cool kids doing that anymore) and someone were to report a bug there, but it seems unlikely.
I’ve noticed that almost every book has 4-5 stars on Goodreads, which makes that system sort of useless. Maybe most grade inflation can be explained by the hypothesis that humans are drawn to a binary. A nuanced five-point scale requires more thought than a good/bad dichotomy. And de gustibus non est disputandum.
I agree. Not sure if can stick to it but I’m planning to just… not read reviews of my game. Happy to discuss it with anyone, of course, but I don’t think reviews will be my business.
Speaking only for myself, of course. What other people do with reviews isn’t my business, either.
e: which isn’t to say I won’t be grateful to people for playing it and talking about it.
I read reviews and I care about them. So far, I’ve only had one that I thought said an unfair (although not mean) thing, but that’s not my call, and I can’t feel resentful about someone else’s opinion. That person (and everybody who writes reviews) actually took the time to play my game and write something about it, and that’s kind of amazing, no matter what they say. I like hearing good things, but I’ve gotten more out of the criticism than the praise, as it’s usually fair. I’ll eventually rack up a good crop of 1 and 2 star ratings, and surely it will sting a bit, but since it’s a forgone conclusion that that will happen, I’m sure I’ll deal.
I’ve said it before but it bears repeating: if you’re going to put your art out publicly, you’re lucky if anyone notices it at all. And if people notice, someone is going to hate it and say mean things, always and every time. It is the Law of Art.
Quick update: even though this review had only 1/3rd “helpful” votes, the author actually responded to it! He said he initially didn’t realize the offending content was an issue, and took it out of the game. I think this reaffirms that (a) new authors are open to changes, and (b) even mostly “not helpful” reviews can still have an impact. I’m not some crusader against AI art, but I’m glad the author read my comment and decided to look into it more on his own volition.
criticism are much more important than stars, the latter can’t convey constructiveness…
an 1-star rating with a detailed constructive criticism, explaining the weak point of the story and coding is much more important for a developer than a 5-star rating without whys. Hence my posture about giving comments alongside rating.
I didn’t downvote your comment/review, but I have to say that I hope there’s no general social stigma attached in IF circles to using these empowering new tools.
I’m not necessarily an all-out defender of AI art, but I don’t think it can be treated as a foregone conclusion that its use is problematic.
The way the exchange comes across to me is that you’ve successfully shamed the author out of using AI art (and it’s still not clear what there is to be ashamed about in the first place). You’ve used the public space of IFDB to give others a “heads-up” (as if AI art was something that needed a trigger warning) instead of just emailing(*) the author to state an opinion of yours, and you say the author “respond[ed] appropriately” when he wrote that he was embarrassed and that he removed the AI-generated images.
It’s not an illegitimate use of IFDB, but yeah, hmm, I’m not enthusiastic about it in this instance.
(*) website → linktree → mail is listed there
(Sorry for getting further off-topic here.)
Also, a tangent about this: “it was immediately recognizable as the “AI art” style to me”: While that specific image is indeed quite recognizable as the style of some current generators, that’s not necessarily so across the board, which opens up the possibility of mistakes and false accusations.
I think sending an email would’ve felt more like a personal accusation or forcing him to “do” something, which wasn’t my intent. I wasn’t imploring the author to change anything, and for me, it feels like he took my review into account and decided to look into the potential problems, erred on the side of caution, and removed them. Making prospective players aware of something controversial (and maybe they won’t even see a problem with it) versus shaming the author to remove it are not the same.
I don’t think there’s a general consensus yet, but the biggest problem is that it can very easily generate plagiarized content – and since you don’t know who the original author is, you can’t credit them. Same with the chance of it recreating copyrighted character designs, which could be from media you aren’t aware of.
I found a Reddit thread where people debate the legality and ethics of AI art, and I know a game called High on Life was underfire recently for using it. Regardless of the side you take, it’s a complicated issue, and there is a high risk involved. From how I’m reading it, the author of OpenCola chose to avoid these problems and make his own art once he was made aware of it.
(And for the “AI art style” comment, after I saw the image in-game, I re-read the description and saw that the game’s art was made in Midjourney. I don’t think mistakes or false accusations will happen because IF games with human-drawn art usually have a more distinct and consistent art style, and will credit the artists respectively.)
Not to stay off-topic but I want to provide a possibility that isn’t being considered, and it puts artists like me at serious risk in a post-“AI” art world.
Not every artist has a distinct and consistent style. Including myself, I happen to know a couple other artists across various websites that use several different art styles. With every image I create, I effectively have to re-discover a style for myself, because I don’t commit one to habit.
So, to avoid getting flagged, are artists like me supposed to add a disclaimer that says “I promise I made my own images in this game; sorry for the inconsistent art style”?
I’m just a bit worried because of how hard it is for me to make visual art at all, and artists on Reddit are now getting removed by witch hunts for coincidentally having a style that “seems to be” generated by an AI.
Mistakes and false accusations will absolutely happen, and are happening. AI styles can be as wide and varied as human styles. The stereotypical “AI style” has just risen to pop culture because it mimics what is considered high-quality artistry.
EDIT: To be clear, I’m not concerned about anyone’s opinion on “AI art” in this specific conversation. I’m concerned that pockets of the Internet have recently started playing The Thing (or Among Us) by calling out potential examples of AI art based on visual style, when the whole point of the technology is that it could look like anything. In Scrooge’s case, the suspicion happened to be correct, and was also confirmed by the author.
I don’t feel the same about a private email versus a public comment (assuming that both would be polite, as your comment is), not least since the latter seems to require more strongly that the addressee reacts and gives an account of himself. And the “heads-up” part seemed to me to indicate something going in the direction of a moral reproach, rather than just a prudential advice for the benefit of the author.
But okay, these are probably issues of perspective about which one can agree to disagree, and just to be clear, I don’t doubt your sincerity or positive motives.
I agree that this is probably the biggest problem facing AI-generated art, although I’d add that it’s not totally unique to that domain. It might also occur if a commissioned artist is cutting corners and copying stuff.
In fact, assets have been pulled from the Unity Asset Store where the asset creators plagiarized existing games and art (manually, without using any AI). Game creators using assets from there (or from Itch, Artstation, Deviantart or anywhere) need to be quite careful.
Maybe (hopefully) in the future, better models will be created on cleared data, and better reverse image search algorithms will be able to weed out those images which are direct rip-offs.
I’m sorry to hear that that’s been happening to you; I’m not an artist myself, just someone who’s keeping up with the AI art news and played a lot of AI Dungeon back in the day. My theories and reading don’t compare to someone who’s actually been in the situation.
But for my two cents, I think the key is that an artist has a vision for what they want to make. They understand how it fits into the game or story they want to tell. AI is simply incapable of doing this. It can generate massively detailed portraits of what are supposed to be the same character a hundred times in a row, but never make them look right; proportions are off (especially faces and hands), it doesn’t understand the character’s personality, backstory, or role, and it doesn’t know the setting to put them in. This inconsistency is something a human can never recreate, and that’s why I feel it’s infeasible for full-fledged works. Human art has any amount of care, thought, and effort put into it. AI art simply grabbed a bunch of stolen pieces and mashed them together.
This was a fascinating article on the subject of AI-generated art. As someone who routinely sells at art fairs, I found that I was torn about my feelings on it. On one hand, using something like Midjourney IS a creative endeavor-- I’ve tried it myself and found it surprisingly… artistic to tweak my request in order to get what I want. On the other hand, I can see why it’s so upsetting to other artists. And on the first hand again, isn’t all digital art related to this process?
Until AI can embroider and cast glass, I’m not worried about it, but I can see why many artists are fretting-- both about AI art itself, and regarding Jess’s concern about overzealous policing of art.
The idea of a multi-dimensional scoring system is intriguing (You probably weren’t actually suggesting that, but it came to mind as I was reading your comments). Authentic critical review is always multi dimensional. (Succeeded in this way, fell short in this other way)
Then again, those distinctions are better brought out in a written review. Awarding a game a star rating (even one star) together with a short written comment, will move it to the top of the “new on ifdb” list, providing some attention to the author for the work. As they say, even bad publicity is publicity.
