That’s not what they said. There is a difference between using AI in a short sighted effort to cut costs and using it to enhance content created by people. AI is a broad term, and just because a bunch of rich asshole morons are misusing a version of it that does have use does not make it automatically bad. AI, Generative or not, is just a tool.
There have been games that have procedural generation for decades in one form or another to create practically infinite content for players, but they are always limited in other ways. Minecraft can generate an “infinite” world, but what you do in the world is limited to what has been ready built. Hell, Games like Skyrim randomly generate NPCs all the time, but they are shallow and don’t really add much to the game.
Having people build out the mechanics, the spells, the world, and other features with a basic foundation of game play and then having AI implemented to combine those features in a way based on player interaction, or create NPCs that are doing similar things the player can that can make the world feel more alive is likely the next real advancement that games will have.
Sure, you could have people make hundreds, if not thousands, of NPCs, but they are going to be very derivative and you’ll see the usual “copy paste” people that aimlessly wonder around or do one or two things and making that many NPCs that aren’t story driven would be mind numbing work.
Sure, you could have people make hundreds, if not thousands, of NPCs, but they are going to be very derivative and you’ll see the usual “copy paste” people that aimlessly wonder around or do one or two things and making that many NPCs that aren’t story driven would be mind numbing work.
If you think “damn I need to make a bunch of fluff here to fill up space but I find the process excruciatingly boring and unfulfilling” please for the love of all that is good and beautiful please stop making art, it isn’t making anybody’s life better including yours. Make art because you desire to create the thing you are actively making in your hands, and if your heart tells you that it isn’t worth it, that means you aren’t making art that is worthwhile.
Procedural generation is a staple of many gaming genres already, but the difference between procedural generation and AI is that a human can ensure that procedural generation will reliably reproduce interesting content, AI has no such proven ability and you can’t just assume that it will attain that ability at some point. Crucially in all the critically successful games that leverage procedural generation the motivation is not to provide endless content but rather to “shuffle” the deck of a carefully hand selected array of cards to create a specific experience that never repeats quite identically which is a crucial element of mechanically challenging roguelike game design.
Enter The Gungeon wouldn’t be made better by swapping out the careful level design considerations for AI generated slop, it would ruin the precisely crafted balance and gamefeel that has lead to it being considered a modern classic.
Procedural generation is not AI, it is in fact philosophically the opposite of AI in that procedural generation procedurally creates and mixes content instead of machine learning which just learns to bullshit pattern match from material that is 9 times out of 10 stolen from exploited artists. One of those things you can tweak to reliably provide fun, challenging and interesting level design that remixes human created elements in ways that don’t undermine the humane element of them and the other is a bullshitting machine. I am sure the bullshitting machine will get better, but it will never not be a bullshitting machine and the success of procedurally generated design in gaming really has NOTHING to do with what we now define as “AI” whatsoever. Rather, on the contrary procedural generated design has far more in common with the now largely forgotten attempts in AI research to create procedural intelligence by explicitly defining thinking and logic routines that could then be modified and built upon by a logical agent operating in a human defined architecture.
If you want to talk about AI in the context of how people used to define AI before the explosive growth and hype of machine learning basically erased an entire branch of research from the public consciousness, well yes that older style of AI design has actually shown itself to be continually relevant to game design…
That’s not what they said. There is a difference between using AI in a short sighted effort to cut costs and using it to enhance content created by people. AI is a broad term, and just because a bunch of rich asshole morons are misusing a version of it that does have use does not make it automatically bad. AI, Generative or not, is just a tool.
There have been games that have procedural generation for decades in one form or another to create practically infinite content for players, but they are always limited in other ways. Minecraft can generate an “infinite” world, but what you do in the world is limited to what has been ready built. Hell, Games like Skyrim randomly generate NPCs all the time, but they are shallow and don’t really add much to the game.
Having people build out the mechanics, the spells, the world, and other features with a basic foundation of game play and then having AI implemented to combine those features in a way based on player interaction, or create NPCs that are doing similar things the player can that can make the world feel more alive is likely the next real advancement that games will have.
Sure, you could have people make hundreds, if not thousands, of NPCs, but they are going to be very derivative and you’ll see the usual “copy paste” people that aimlessly wonder around or do one or two things and making that many NPCs that aren’t story driven would be mind numbing work.
If you think “damn I need to make a bunch of fluff here to fill up space but I find the process excruciatingly boring and unfulfilling” please for the love of all that is good and beautiful please stop making art, it isn’t making anybody’s life better including yours. Make art because you desire to create the thing you are actively making in your hands, and if your heart tells you that it isn’t worth it, that means you aren’t making art that is worthwhile.
Procedural generation is a staple of many gaming genres already, but the difference between procedural generation and AI is that a human can ensure that procedural generation will reliably reproduce interesting content, AI has no such proven ability and you can’t just assume that it will attain that ability at some point. Crucially in all the critically successful games that leverage procedural generation the motivation is not to provide endless content but rather to “shuffle” the deck of a carefully hand selected array of cards to create a specific experience that never repeats quite identically which is a crucial element of mechanically challenging roguelike game design.
Enter The Gungeon wouldn’t be made better by swapping out the careful level design considerations for AI generated slop, it would ruin the precisely crafted balance and gamefeel that has lead to it being considered a modern classic.
Procedural generation is not AI, it is in fact philosophically the opposite of AI in that procedural generation procedurally creates and mixes content instead of machine learning which just learns to bullshit pattern match from material that is 9 times out of 10 stolen from exploited artists. One of those things you can tweak to reliably provide fun, challenging and interesting level design that remixes human created elements in ways that don’t undermine the humane element of them and the other is a bullshitting machine. I am sure the bullshitting machine will get better, but it will never not be a bullshitting machine and the success of procedurally generated design in gaming really has NOTHING to do with what we now define as “AI” whatsoever. Rather, on the contrary procedural generated design has far more in common with the now largely forgotten attempts in AI research to create procedural intelligence by explicitly defining thinking and logic routines that could then be modified and built upon by a logical agent operating in a human defined architecture.
If you want to talk about AI in the context of how people used to define AI before the explosive growth and hype of machine learning basically erased an entire branch of research from the public consciousness, well yes that older style of AI design has actually shown itself to be continually relevant to game design…