sometimes??
Jennell Jaquays died, she was only 67. She seemed like one of the good people from the era of the original dungeon makers. Her art was in the AD&D books, she worked on various computer games, and i think she made a large contribution to RPG adventure design. Hereās her illustration of Ningauble of the Seven Eyes:
She did a lot of level design on Quake 2 and 3 too!!
New Momodora is out today
IDK, I think AI can be used in really effective ways that would just help. Like using AI for āinbetweenersā on art, or to mock up color variants or slight costume alterations.
what does this even mean
what the hell is a traditional gamer
why is new gamers having different tastes laughable instead of just being different and valid
Iām old enough and stupid enough that I feel like Iāve been seeing this same thing every few years for the past few decades
and yet it doesnāt make me feel any less incredulous about it when it pops back up
trad-gamer wife
idk im usually pretty cynical about things but i think if people find it to be an extremely important distinction there are going to be a lot of game developers who continue to not use ai stuff, i mean if there are still people shooting 35mm film on full manual slrs from before autofocus was invented and developing it at home, or people playing live instruments and stuff, idk
my feelings about AI in games is i do think it will kind of suck and will take away jobs, however there is some part of me that is vaguely excited by the idea for myself to think about trying game dev out without necessarily needing all of the steps to begin. i enjoyed how Pico-8 lets you look at the code and you can sort of tweak it and begin to figure out what is going on with someone elseās project, and perhaps in that way, AI could create something of a template that a user could then spend time tweaking and learning from and perfecting into something that is actually interesting.
like the gap between understanding modular synthesizers versus analog versus digital, etc. it is most helpful to know them from the beginning, but if you just want to dive in and start making things that sound coherent, there are options
AI could create something of a template that a user could then spend time tweaking and learning from and perfecting into something that is actually interesti
my brain will melt out my ears if i have to navigate ai discourse but i want to repost this excellent article from ted chiang on the subject of llms because it has a pretty elegant explanation of why llms are not and will never be useful for this sort of thing
the whole thing is worth a read but the most relevant bit is this:
Your first draft isnāt an unoriginal idea expressed clearly; itās an original idea expressed poorly, and it is accompanied by your amorphous dissatisfaction, your awareness of the distance between what it says and what you want it to say. Thatās what directs you during rewriting, and thatās one of the things lacking when you start with text generated by an A.I.
i keep coming back to this when people talk about their hopes for the technology, this idea that somehow it will remove the need for that first step of putting words on paper (either literally or a base to work off and refine in another medium), but if you remove the creative process from your art whats left?
making games is more approachable than ever, and i think especially the tools around codeless game dev will continue to improve without the need for llms
you can already download a template for a ball rolling game, an fps, a platformer, sprites and models, mishmash them together and see what comes out, download rpgmaker, put some words in an npcs mouth and delight yourself as they read your text back to you just⦠please make it your own
i mean i have thoughts, but i donāt wanna make anyoneās brain melt out of their head, so iāll keep them to myself
the one thing i will say about AI is i bought some mastering plugin deeply on sale (Izotope Ozone) that has an AI assistant that will listen to your track and make suggested tweaks based on that. and it is like, tremendously valuable for me as a person who makes music and does not understand/is not interested in mastering and doesnāt have the money to pay people to master. itās honestly a very acceptable substitute from a very reputable company that made me feel more comfortable about what iām putting out. if i had a lot more money i probably would pay someone to master, but thatās not an option right now. so thatās the kind of thing i imagine AI being very useful for. not the end product as much as just a creative assist to help you shape the end product. and there will always be people who are allergic to any of that stuff.
yeah i mean this was thing people were experiencing on the onset of casual games/phone games at the end of the 00ās. a lot of people thought that stuff was going to take over the industry, and it didnāt. mobile games are obviously still huge but have become this weird gray area quarantined off from other game development and owned by various entrenched mobile game companies. Roblox and Fortnite are especially huge but thereās no guarantee theyāll have the same mass appeal to kids 10 years from now that they do now. if anything it seems to me like a lot more younger people are extra invested in things like Resident Evil, Silent Hill, etc than i ever remember people being. Fortnite and Roblox are like digital places they go to - but those other games are things theyāre more invested in as works. and they feel more eternal and less flavor of the month than they used to.
i think whatās alienating to me now just from a full-on consoomer perspective more than anything is new stuff in genres i have actual nostalgia for (i.e. retro FPS games, 3D platformers, immersive sims, Silent Hill-style horror games etc) are being made at a rate they werenāt before⦠and iām not as into most of it as i thought i would be. i feel a little bit like thereās a backwards-looking zeitgeist to the 90ās and early 00ās right now. which makes sense, but thereās definitely a diminishing returns feeling to a lot of what comes out of it for me. a lot of that is the current marketplace and the way games are made now (i.e. made cheaply by amateur developers for niche audiences, and thereās an emphasis on like aspiring to be a ārealā game developer and look like a respectable entry into a genre while also not challenging those audiences expectations too much so they donāt get mad at you). i kinda felt that way about a lot of big 2D platformers too tho. no one ever sold me on why Shovel Knight or whatever was āimportantā either lol.
i think that was more of a caveat that contentdeleted isnāt interested in debate and just wanted to share than a shut-out. like ellaguro raised, izotope neutron/ozone is a great example that doesnāt replace the blank page, more like replacing your copyeditor. whatās the game development equivalent of that?
yeah for sure, and i didnt mean to call you out specifically either @isfet, just wanted to respond to what i feel is a particularly common sentiment
if i imagine somehow using a more refined version of something like copilot for end to end and regression testing, i can absolutely see the value in that at a point in the development cycle where you know what something should be, and just need to make sure you havent broken anything with your latest changes
in my mind i imagine something more assistive, and i understand the nuance and artistry of coding and really learning it, but as a 40-year-old adult who has almost no real free time (or energy) to dedicate to really learning to code, i feel like having something where i could be like āapproximate Ninja Gaidenā and then further ask some kind of AI to āisolate the hit-detection code for (enemy)ā or what have you, this would save me a lot of hours.
this is perhaps fantastical and implausible and maybe speaks to my own ignorance on certain subjects or not really spending time discussing AI in any meaningful way outside of just observing what folks who know more or are directly affected by it are saying.
and again, i understand the nuance between the example iām about to give, but when i read interviews with game devs from the 80s and 90s, a lot of times these folks were just studying (stolen?) code from other companies and figuring out how to replicate or improve or modify it for whatever project they were working on (most recently, the Ninja Gaiden director was talking about studying how Castlevania stored sprites to use in NG). i get that having the baseline knowledge of coding makes this entirely different (like studying the way a painter paints or an electronic musician mixes), but i think that in spite of things getting a lot more accessible than they used to be, there is still no coding equivalent of say, picking up a guitar and fucking around until you start making sounds you like, etc.
this makes sense to me
this does not, this feels like itās eliding every interesting choice you can possibly make en route, not to mention a lot of the interesting conversations around scale or whether we actually need or benefit from having AI to generate everyoneās own private ninja idaho, as opposed to having shared works that we build culture around
yeah maybe itās poorly thought out. maybe itās more like, extracting the code of something i admire, and then asking it to study that for me and put what is happening in āplain English,ā because sometimes when i look at advanced or intermediate code, as a beginner, i donāt necessarily understand why something is working the way it does, or why it doesnāt work how i expect it to. getting real time feedback to explain it in a more verbal way would be particularly help to me, specifically
scientific coding (as opposed to business coding) tends in this direction. in languages like python or ruby i like to draft code in a breakpoint using an interactive terminal (repl) and then paste it into the source file once iām happy with it. jupyter notebooks are a more elaborate and persistent version of this way of coding
it mostly pertains to massaging data structures or what have you, using an approach like that to gradually rough-in a game would be nice but i imagine thereās a very high floor of complexity there, youād have to lay a lot of groundwork before seeing any meaningful results? iām not a game dev though
Iām dashing this out on break so apologies for being behind on the thread but the bummer is music has a live creation element. Using AI in video games or graphic art in general is more like using weird ingredients when cooking. If you know what to look for you know but most folks might be non the wiser because it all happens behind the scenes. Hence why food ingredient labels is something we legislate. I canāt imagine AI art warning labels being required by law (but it was something addressed in those Hollywood contract negotiations so maybe thereās hope)
thanks, this is a great article. also you probably owe all of my current and future students an apology because iām going to find some way to make them read it.