Yeah, that was my understanding of things too. What I’m curious about is how the Indie Game awards define it. Because if games that use ((Procedural Generation) AND NOT (Generative AI)) are permitted, then that would surely require a way of cleanly delineating between Generative AI and the rest of procedural generation that exists beyond generative AI
I’m not so much talking about machine learning being implemented in the final game, but rather used in the development process.
For example, if I were to attempt a naive implementation of procedurally generated terrains, I imagine I’d use noise functions to create variety (which I wouldn’t consider to be machine learning). However, I would expect that this would end up producing predictable results, so to avoid that, I could try chucking in a bunch of real world terrain data, and that starts getting into machine learning.
A different, less specific example I can imagine a workflow for is reinforcement learning. Like if the developer writes code that effectively says "give me terrain that is [a variety of different parameters], then when the system produces that for them, they go “hmm, not quite. Needs more [thing]”. This iterative process could, of course, be done without any machine learning, if the dev was tuning the parameters themselves at each stage, but it seems plausible to me that it could use machine learning (which would involve tuning model hyperparameters rather than parameters).
You make a good point about procedural generation at runtime, and I agree that this seems unlikely to be viable. However, I’d be surprised if it wasn’t used in the development process though in at least some cases. I’ll give a couple of hypothetical examples using real games, though I emphasise that I do not have grounds to believe that either of these games used machine learning during development, and that this is just a hypothetical pondering.
For instance, in Valheim, maps are procedurally generated. In the meadows biome, you can find raspberry bushes. Another feature of the meadows biome is that it occasionally has large clearings that are devoid of trees, and around the edges of these clearings, there is usually a higher rate of raspberry bushes. When I played, I wondered why this was the case — was it a deliberate design decision, or just an artifact of how the procedural generation works? Through machine learning, it could in theory, be both of these things — the devs could tune the hyperparameters a particular way, and then notice that the output results in raspberry bushes being more likely to occur in clusters on the edge of clearings, which they like. This kind of process would require any machine learning to be running at runtime
Another example game is Deep Rock Galactic. I really like the level generation it uses. The biomes are diverse and interesting, and despite having hundreds of hours in the game, there are very few instances that I can remember seeing the level generation being broken in some way — the vast majority of environments appear plausible and natural, which is impressive given the large number of game objects and terrain. The level generation code that runs each time a new map is generated has a heckton of different parameters and constraints that enable these varied and non-broken levels, and there’s certainly no machine learning being used at runtime here, but I can plausibly imagine machine learning being useful in the development process, for figuring out which parameters and constraints were the most important ones (especially because too many will cause excessive load times for players, so reducing that down would be useful).
Machine learning certainly wouldn’t be necessary in either of these examples, but it could be something that could make certain parts of development easier.
Can someone help me to understand the difference between Generative AI and procedural generation (which isn’t something that’s relevant for Expedition 33, but I’m talking about in general).
Like, I tend to use the term “machine learning” for the legit stuff that has existed for years in various forms, and “AI” for the hype propelled slop machines. Most of the time, the distinction between these two terms is pretty clean, but this area seems to be a bit blurry.
I might be wrong, because I’ve only worked with machine learning in a biochemistry context, but it seems likely that modern procedural generation in games is probably going to use some amount of machine learning? In which case, would a developer need to declare usage of that? That feels to me like it’s not what the spirit of the rule is calling for, but I’m not sure
Ooh, Barnsley! That’s actually super close to where I hail from (we had a Barnsley phone number and post code, despite not technically being in Barnsley). That’s so cool, it’s not the kind of place you typically see depicted in media
“Northern Englishest”
As a Yorkshire lass living in Manchester, you have me intrigued with this description. Hell, I’m more than intrigued — you’ve sold me on it
I’ve not played Factorio Space Age yet, but I’m looking forward to it whenever I next get a hankering for Factorio.
Your pitch has sold me on it. Yet another game to add to my wishlist
That sounds like a space version of Eco, with the roles stuff. In Eco, it’s impossible for one person to acquire all skills, so people on a server have to specialise.
I started out as a miner, to honour my late best friend who was a dwarf at heart and would definitely have been a miner if he’d been playing with us. Then I branched out into masonry to make use of the absurd amounts of stone I’d been mining. If I wanted something made of wood, I had to go flutter my eyelashes at my friend who had started out as a logger and branched into carpentry. I enjoyed having a domain that was my own, and a clear way to be useful to the server. Other players had some level of mining and masonry skill by the midgame, but for anything serious, they had to wait until I was online.
It sounds like Space Station 14 is far more hectic than this, but in an interesting way. I wonder if it will scratch the same itch that Eco did wrt being useful in a clear role
Space Station 14 sounds interesting. What kind of multiplayer is it? I.e. is it one where the typical experience is to play with randoms via matchmaking, or is it a game best enjoyed with friends?
I have discord server full of nerds who I played games with during COVID (and its aftermath), and this might be a good excuse to see if I can reawaken that server for games
You’ve reminded me that I still need to finish that. When I started it, I played it so much that I burnt myself out on it a tad (not in a bad way, just in a way that requires I take a break and play something else for a while). I’m looking forward to getting back to it.
I didn’t play the first game, but I remember seeing a lot of the promo/development stuff about it because my partner at the time was super interested in it. My impression of the first game was that it was ambitious and interesting, but rocky in its implementation, but the second one is a refinement in all the ways you would expect a sequel to be. Certainly I have enjoyed it thus far
Edit: Steam tells me that I have 133.5 hours in this game, bloody hell. In my original post, I mentioned that I expect that the actual data in the Steam year-in-review will differ from what I remember of 2025, and this appears to be a great example of it. It seems like this was one of the games that completely dominated the first half of 2025 for me, and I didn’t even remember it
Thanks for replying and giving me yet another game that I’ve not even heard of that I’m probably going to check out.
I’m not a huge RTS person, but occasionally I get a strong craving for one. Next time I do, I’ll see if Beyond all reason scratches that itch
I’ve never heard of Drova or Cruelty Squad, so thanks for the recommendations. This thread has given me so many interesting games to check out, thanks for replying
If @[email protected] hadn’t already sold me on Cruelty Squad, you certainly have now. In terms of vibes, it sounds right up my alley.
And I do love a bonkers community. I find that when I get into a piece of media (whether that be a game, TV series or something else), I really enjoy participating in what I call “fandom tourism”. I enjoy dipping my toe into the community after I’ve engaged with the media itself, and it feels like bonus content. I don’t tend to stick around in any fandoms, so that means that even if a community is bonkers in a bad way (e.g. lots of drama), I even sort of enjoy being able to understand and spectate those dynamics, as a quasi-outsider
I’ve heard so many good things about Lies of P that I think I’ve been avoiding it in a similar way to how I was irrationally reluctant to play Hollow Knight. It’s a bit of a moot point at the moment, because I don’t currently have the brain space to get my teeth into a Soulslike, but when I do, I should resist that silly instinct of mine.
I’ve not heard much of Dispatch, I should check it out
Most of those games are ones I’ve never heard of before, but you’ve really sold me on them, especially Split Fiction and UFO 50
(Mini tangent, but I find it interesting how, in this age of algorithmically driven slip content, I cherish the opportunity to find little snippets of meaningful connection with my fellow humans. Like, I don’t know you, or anything really about your preferences or tastes in games, so what reason is there to put much weight in your recommendations? You’re just a random person on the internet, after all. But no, your recommendations feel meaningful because you’re a person who cared enough about these things to write about them, and matters to me (especially in our current climate))
If I was going to try out Split Fiction and UFO 50, which would you recommend I start with?
Nice! I haven’t attempted Sekiro yet, but it’s high up on my list. I am saving it for when I have the brain space to take a proper crack at the game. I remember that my first exposure to Fromsoft games was in 2017, when I attempted Dark Souls 3 during a Summer where I extremely burnt out due to doing a soul-sucking internship. I bounced off of it so hard, and that taught me that I need to be in the right headspace to play certain games.
Despite the high skill level required, I actually found that it was quite forgiving for people who were learning. I barely did any parrying until I was well into Act 3, for example. I like the way that the feedback for dodges work — I started trying to parry more when I realised that I was consistently getting perfect dodges, which meant that if I had parried, it would have been successful.
I also like the way the difficulty works in the open world. It reminds me of games like Fallout: New Vegas, where the enemies aren’t scaled to player level, so you can be dumb/brave and wade into encounters that are way beyond your power level. Sometimes that works out surprisingly well, but often you try fighting a difficult enemy and get pwned so thoroughly that you accept that you’ll have to come back later. In Expedition 33 especially, it is super viable to just go and explore elsewhere and come back with more levels, better weapons and better pictos. The beautiful world also means that exploring is fun even without the mechanical perks.
I don’t use Arch, but I am eternally grateful for their excellent documentation.
I am also grateful to you for your comment, because this is a good idea
I agree with the ethical standpoint of banning Generative AI on the grounds that it’s trained on stolen artist data, but I’m not sure how tenable “trained on stolen artist data” is as a technical definition of what is not acceptable.
For example, if a model were trained exclusively on licensed works and data, would this be permissible? Intuitively, I’d still consider that to be Generative AI (though this might be a moot point, because the one thing I agree with the tech giants on is that it’s impractical to train Generative AI systems on licensed data because of the gargantuan amounts of training data required)
Perhaps it’s foolish of me to even attempt to pin down definitions in this way, but given how tech oligarchs often use terms in slippery and misleading ways, I’ve found it useful to try pin terms down where possible