Comment by everdrive
4 hours ago
Will AI games be good? I think they could be, but in practice when you reduce the friction what happens is that the signal to noise ratio gets much worse. The market will be flooded with garbage, and so per capita games will become worse, even if the absolute number of good games rises. So, I'm not excited. We already have too many games. The market is flooded and fragmented. There are more games released now than I have remaining hours in my life.
> Will AI games be good? [...] The market will be flooded with garbage, and so per capita games will become worse, even if the absolute number of good games rises.
Funny you use "will be" here, because this is already happening. Have you looked at daily Steam releases lately? It's 90% AI-generated hentai games, asset flips, or AI-generated hentai asset flip games.
Right now the limitations of genAI tech are limiting the types of games that can be pumped out, but Genie is just going to be more of the same; instead of a clearly AI-generated, painfully uninteresting hentai game, you're going to get clearly AI-generated, painfully uninteresting Call of Duty clones. You can ask the developers of Concord, Highguard, Marathon, etc. how successful bland shooters are these days.
I wasn't aware of this, and this is an important comment. I think the imaginations of the people on HN are really turning with what could be possible in principle. In practice, it looks just like you've described; low quality engagement-focused garbage, just like your facebook and youtube feeds.
Gaming is already flooded with garbage. Pretty much on every level. From cheapest possible shovelware sold for cents. To soulless derivative attempts in live service games. And anything in between. It can probably get worse, but I doubt anything will actually change.
It is definitely a new frontier. How do you review a game that isn't reproducible? Similarly how do you do play testing? You don't get the same "shared experience" outside of multiplayer and people experiencing it together. It is taking current generation procedurally generated games to the next step, but where do you find a threshold of "this blob of constraints is _the_ game"?
I'm not saying that video games should be confined to today's paradigms, I'm just happily curious to see what will happen with it.
From an art perspective, AI seems to be quite a ways away from someone being able to make a coherent general game.
The self-consistency just doesn't seem to be there to make a coherent playable game (easily, broadly - many types of games, styles, etc).
Music gen AI is pretty good, and with a reasonable amount of finagling, you might be able to pull that off generally.
From a coding perspective, you can probably make simple games with minimal skills already, and I can definitely see it being possible to create pretty sophisticated games with minimal bugginess with minimal coding skills in the near future.
Is that combination good enough? I doubt it outside of niche games for quite some time
Yes, I've played a lot of indie hobby-tier niche games and I will say that AI assets are a big help for solo hobby devs who wouldn't otherwise be able to finish the game. At the same time, the AI art is noticeably inconsistent in an 'uncanny valley' way that breaks immersion, making the game quality poor.
> Music gen AI is pretty good
Is it? I have quite a few friends that are relatively successful music producers/djs, and they don't seem impressed. A lot of music gen AI fails to match prompts at all. Maybe I am using the wrong tools.
It also seems quite difficult to iterate on a sound.
Are you talking specifically about Genie?
Sounds like an increased demand for game critics. Or perhaps the same thing that has happened with web search will apply (at least in my experience)--less clicking on search results, more going directly to trusted sites. Less trying out new studios and IPs, more gamer brand consciousness and loyalty?
That's a likely adaptation, yes, but there is no real market demand for a higher volume of (mostly) lower quality games. So will we all adapt, yes, but much like web search it would have been better if it were never ruined in the first place.