Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Building an AI game studio: what we've learned so far

braindump.me

Post reply on HN
Print view View original

181–190 of 312 posts

Re: Building an AI game studio: what we've learned so far

#181

Post by viking123 »

jsheard wrote: post #75
bogwog wrote: post #30

They mention using a service called Meshy for generating models, and I noticed that they have a free plan where the generated assets have a "license" of CC-BY-4.0[1]. Any lawyers in here that can comment on that? They're claiming copyright ownership over models generated from customer prompts, and I doubt they trained their model exclusively on stuff they own/licensed. 1: https://www.meshy.ai/pricing

https://app.meshy.ai/preview?web=018f8ef6-e562-7e19-b0d7-5f6... https://app.meshy.ai/preview?web=018f4d04-6a56-701e-9abc-62a... IANAL, but I think Blizzard and Marvel would beg to differ that these are kosher for commercial use. They're also very poor quality models but that's a separate issue.

I like how the first one has two faces, one on the front and one on hair at the top

Re: Building an AI game studio: what we've learned so far

#182

Post by Hammershaft »

somenameforme wrote: post #143
dwallin wrote: post #29

Earlier quoted context omitted.

Making an excellent game isn't even enough these days, that's just table stakes. With so many games out there it's a market of lemons. You need to have a way of rising above the noise. For AAA, it's having budgets to make games on a bigger scale than everyone else. For mobile games, it's generally about leveraging a large war chest to fuel an advertising-driven margin game. For prestige/indie games you need something…

An excellent game is, practically by definition, a game that people would recommend. And the Steam algorithm does a pretty good job of snowballing success. My favorite example here is Kenshi. [1] It started with modest interest and within a few months it was down to literally tens of players. This is normally where somebody driven by profit would just shrug, complain, and go try something new. But they guy clearly be…

For every Kenshi there are many dozens of excellent indie games that languish. The steam algorithm is not as cooperative for surfacing small niche games as it was of Kenshi, and it's harder to get your game seen in a more saturated market. There has been a gradual shift from the 2010's where nearly all great games we're handsomely rewarded.

Re: Building an AI game studio: what we've learned so far

#183

Post by rurp »

ElFitz wrote: post #90
pjc50 wrote: post #23

Earlier quoted context omitted.

Itch.io claim to host over 200k free games. AI makes standing out harder; we're not ready for the sludge era of having seven billion people and ten billion games.

Except that’ll just be a fluke. The next thing isn’t a never-ending list of AI generated games stored somewhere that we’ll have to sift through somehow. The next big thing will be ephemeral games, created on the fly based on the context and the preferences expressed, both explicitly and implicitly, by the player’s (or players’) past behaviour (and, for multiplayer games, that of their social graph). That and perhaps,…

I think this will describe a category of games, assuming the tech eventually gets good enough, but I can't imagine this replacing all games. Most people won't want to craft a 10,000 word prompt to layout the kind of game they want to play.

There will still be professional developers who figure out the gameplay mechanics, design style, difficulty balance, progression, and more. Maybe there will be more customization around the edges due to AI, but most players want to grab a ready to go game and get playing.

Re: Building an AI game studio: what we've learned so far

#184

Post by jsheard »

viking123 wrote: post #181
jsheard wrote: post #75

Earlier quoted context omitted.

https://app.meshy.ai/preview?web=018f8ef6-e562-7e19-b0d7-5f6... https://app.meshy.ai/preview?web=018f4d04-6a56-701e-9abc-62a... IANAL, but I think Blizzard and Marvel would beg to differ that these are kosher for commercial use. They're also very poor quality models but that's a separate issue.

I like how the first one has two faces, one on the front and one on hair at the top

Oh god I didn't even notice that. That's one of their "featured" showcase models, too.

Re: Building an AI game studio: what we've learned so far

#185

Post by PurpleRamen »

skyyler wrote: post #127
kyleamazza wrote: post #89

Earlier quoted context omitted.

But we're in this universe. Based on the number of years it took to develop the game, it wouldn't be considered one that achieved "15 minutes of fame". 100 reviews is quite a far cry from your example of Balatro with 33k. I'm not sure where you plan to draw the line on "success", but I think this is a broadly reasonable contrast.

3 of my steam friends have already purchased it. One of them has it on wishlist. I only have 40 steam friends.

I own this game too, got it probably from a bundle-sale, so I didn't really bought it on purpose for the game itself. It also seems to haven been given away for free some years ago.

Re: Building an AI game studio: what we've learned so far

#186

Post by jprete »

fabiopolimeni wrote: post #180
Bjorkbat wrote: post #156

I think a part of the reason why I'm kind of dismissive of AI when it comes to creative endeavors is that I haven't seen it do something that isn't already possible using a "not-intelligent" solution. For example, AI website builders. It's impressive as a tech demo that you can prompt an AI to make a website for you, but otherwise I don't think this is going to be disruptive to the web design industry since it's alre…

Uhm…, even you use something low-code, or even no-code tool, you still have to learn how to code (former) or learn how to use some kind of visual scripting tool (latter). If you could simply type in plain English what are the mechanics you have in mind instead, then this new (simpler) way of creating game logic it may has the potential to unleash creativity from a broader audience. It is like saying why would I need…

If the AI doesn't get the mechanics right, and you have subtle game-breaking bugs that make it impossible to complete, you will never be able to fix it.

Re: Building an AI game studio: what we've learned so far

#187

Post by ElFitz »

user432678 wrote: post #137
ElFitz wrote: post #90

Earlier quoted context omitted.

Except that’ll just be a fluke. The next thing isn’t a never-ending list of AI generated games stored somewhere that we’ll have to sift through somehow. The next big thing will be ephemeral games, created on the fly based on the context and the preferences expressed, both explicitly and implicitly, by the player’s (or players’) past behaviour (and, for multiplayer games, that of their social graph). That and perhaps,…

I wouldn’t be overly optimistic on what an average player can express as a preference. Besides, as a gamer I want to be surprised and entertained, I don’t want to design rules and game mechanics, iterate over balancing the game and finding consistent art style. Unless you meant like really distant future.

I meant something akin to recommendation engines like TikTok’s, applied to generative models instead of user-generated content.

Start-off with generic, "one size fits all", generated content, using a few parameters such as rough area, time of day, time of the wear, what’s currently popular at that time in that area, what’s popular with the person who created the link if they got in through that, etc.

Then, use the player’s actions (playtime, engagement, sharing, whatever) to infer preferences, and fine-tune the model using both context (time of day, location, other players if multiplayer, etc) and those preferences.

Again, recommendation engine, applied to generative models.

I wouldn’t wager on how distant or close that future will be.

I don’t think it’s doable today, but it’s so obvious that I am certain people are currently working on it. And perhaps some duct tape (sorry, heuristics and expert systems) may be able to somewhat fill the gap in current or near-future models’ capabilities.

But I’m certain (unless copyright laws end up killing such use cases), that it’s the future for both streaming platforms (Netflix, Disney, Spotify & all), content-driven "social" media platforms (TikTok, YouTube, Instagram,…), and video games. Most of them anyway. Some might doubt it can rival with the greatest movies, books, or games, but it doesn’t need to. It just needs to be as good as the cheap filler content these platforms are mostly filled with.

I just hope we’ll find a way to preserve those industries so we can keep getting great new things.

Re: Building an AI game studio: what we've learned so far

#188

Post by iamleppert »

This is genius! I currently work on attention loop games (the kinds you see ads for on your social newsfeed). A big bottleneck is the number of new games we are able to produce per week. New games generally have a better ROI and get more installs and clicks than older games, which age out.

Having the ability to programmatically generate new games would be amazing. It would be great to have the game completely generated by AI, and then pipe the screenshots into a video AI tool to create the ads.

Will it support slot based games and gambling card games? My head is just buzzing at the possibilities and the opportunity for profits from this!

Re: Building an AI game studio: what we've learned so far

#189

Post by hiatus »

Hammershaft wrote: post #182
somenameforme wrote: post #143

Earlier quoted context omitted.

An excellent game is, practically by definition, a game that people would recommend. And the Steam algorithm does a pretty good job of snowballing success. My favorite example here is Kenshi. [1] It started with modest interest and within a few months it was down to literally tens of players. This is normally where somebody driven by profit would just shrug, complain, and go try something new. But they guy clearly be…

For every Kenshi there are many dozens of excellent indie games that languish. The steam algorithm is not as cooperative for surfacing small niche games as it was of Kenshi, and it's harder to get your game seen in a more saturated market. There has been a gradual shift from the 2010's where nearly all great games we're handsomely rewarded.

> For every Kenshi there are many dozens of excellent indie games that languish.

Can you list some here?

Re: Building an AI game studio: what we've learned so far

#190

Post by rurp »

password54321 wrote: post #44
kibwen wrote: post #41

Earlier quoted context omitted.

This is subject to survivorship bias. Balatro is a great game that deserves its success. But it's easy to imagine a universe where it simply never got any traction, anywhere, and died in obscurity. Conversely, this suggests that in our current universe, many games just as great as Balatro have quietly passed without ever getting their 15 minutes of fame. At the end of the day, much of success just hinges on luck. The…

I immediately knew someone would claim "survivorship bias" but at the same I knew that they would fail to come up with a single counterexample.

Survivorship bias is the norm in any system with high variance and a low success rate. I would say that arguing against the phenomenon is by far the more out there position.

A analgous statement like, "There are actors with world class talent who never became successful" is one that pretty much everyone would intuitively agree with, outside of maybe an internet agrument. Whether or not I can come up with some talented obscure actors on the fly doesn't disprove the statement.

Post reply on HN