Valve have now confirmed their rules for game developers using AI generated content on Steam, and while they're not banning AI they are going to ensure it's clearly stated for players. Like it or not, "AI" use is only going to increase in the games industry and so Valve had to do something and what they've announced sounds reasonable enough.
When developers fill out a survey for Valve to get their game on Steam, it now includes AI disclosures developers have to fill out and they separate it into two categories:
- Pre-Generated: Any kind of content (art/code/sound/etc) created with the help of AI tools during development. Under the Steam Distribution Agreement, you promise Valve that your game will not include illegal or infringing content, and that your game will be consistent with your marketing materials. In our pre-release review, we will evaluate the output of AI generated content in your game the same way we evaluate all non-AI content - including a check that your game meets those promises.
- Live-Generated: Any kind of content created with the help of AI tools while the game is running. In addition to following the same rules as Pre-Generated AI content, this comes with an additional requirement: in the Content Survey, you'll need to tell us what kind of guardrails you're putting on your AI to ensure it's not generating illegal content.
Valve said they will also "include much of your disclosure on the Steam store page for your game, so customers can also understand how the game uses AI". So with that, if you plan to avoid AI games, at least Valve will give you a clear way to spot them.
On top of that Valve said they're implementing a new system to allow players to "report illegal content inside games that contain Live-Generated AI content", which you'll do via the Steam Overlay.
Together these rules and features allow Valve to be "much more open to releasing games using AI technology on Steam" but Valve said clearly that Adult Only Sexual Content created with Live-Generated AI is not currently allowed on Steam. That one feels like an obvious one to not allow, for reasons I'm sure I don't need to go into.
What do you think to Valve's stance on this?
See the full announcement here.
Whether it will be good for games remains to be seen. When used cleverly, it might help development or add some interesting things to NPC interactions. But I doubt it's going to bring us anything we haven't seen before. My biggest worry would be that it's going to be used as an excuse for an always online requirement and it just turns into a form of DRM.
QuoteWhat do you think to Valve's stance on this?
It's a very Valve solution. Limit their own liability while "providing value to customers" in their habitual phrasing - customers that want AI can have it and customers that don't can avoid it. Developers get to do what they want as long as they provide sufficient data to Valve.
One thing that could provide a wrinkle is language support. If the data sets the AI is trained on are majority English then the output will be majority English, and Valve have recently been tightening up on half-arsed localisation - promo artwork has to be available in all the store page languages, for example.
Last edited by CatKiller on 10 January 2024 at 11:06 am UTC
QuoteLike it or not, "AI" use is only going to increase in the games industryI'm going with not.
AI enables people who are creative but don't give a shite about coding. Just like one can sing beautifully without knowing eff all about what notes and structure there might be on paper.
I think many old schooling gatekeepers will be surprised what creative people can come up with using AI as tool. Like a synthesizer in music maybe. The creative landscape is getting larger. It's the same type of people enabled on different ground who always found math uninteresting in life but are great writing touching poetry.
Maybe we are heading towards a kind of Star Trek world where we talk to computer for instance and get stuff i remember the Linux creator too already said long time ago that keyboards are very bad for human communication ..
Quoting: EhvisI'm all for the disclosure part that Valve has always done on the Steam store (I don't think any other store requires the publishers to specify their DRM restrictions).Unfortunately, this doesn't include the DRM wrapper Valve offers, which is something you need to test yourself. Makes it harder to tag games on VNDB
But since you're asking, DLsite does a great job of this. The DRM is labeled front-and-center, in a big green box, at the top of the store page. It includes information about the DRM, screenshots of how it works and how to use it, and the DRM name. They also require their own PlayDRM implementation to be mentioned (which does work in Wine). If the store page doesn't have any mention of DRM, it's DRM-free. You can also sort by DRM-free games.
DMM does similar but hides it right at the bottom (near the "Specs" section), isn't as clear about what exactly it is, and sometimes doesn't seem to mention it at all.
A popular method of using (what's commonly called) AI is for sketching out ideas and getting rapid feedback on them. If for example, a writer puts in a page worth of text and asks the AI to give its feedback on, say, the themes or tone of the text, that now has to be tagged as AI generated, despite not a single word of the final product being AI generated.
For developers, it's also fairly common to use AI as a talking rubber duck (i.e. bounce human-generated ideas off of it to see if it spots any holes in the logic). Again, none of the code is written by AI, but it would now be tagged as if it was.
If this leads to practically every game having the AI tag, the point of it seems a bit lost.
Quoting: CybolicI'm for the idea, but the language of "content [...] created with the help of AI tools" might cause issues.Really? People trust the feedback of those things? On stuff like themes? That's, um, the word that keeps coming to my head is "pathetic".
A popular method of using (what's commonly called) AI is for sketching out ideas and getting rapid feedback on them. If for example, a writer puts in a page worth of text and asks the AI to give its feedback on, say, the themes or tone of the text
Quoting: Purple Library GuyQuoting: CybolicI'm for the idea, but the language of "content [...] created with the help of AI tools" might cause issues.Really? People trust the feedback of those things? On stuff like themes? That's, um, the word that keeps coming to my head is "pathetic".
A popular method of using (what's commonly called) AI is for sketching out ideas and getting rapid feedback on them. If for example, a writer puts in a page worth of text and asks the AI to give its feedback on, say, the themes or tone of the text
Out of curiosity, what would you consider a non-pathetic way to get feedback?
Last edited by Cloversheen on 11 January 2024 at 5:45 am UTC
See more from me