The Slippery Slope of AI in Games | Semi-Ramblomatic
Added 2025-07-24 15:00:42 +0000 UTCThis week's episode of Semi-Ramblomatic is now available!
Comments
I think this is at it's core a disingenuous argument. You are conflating the advance of tooling to support conventional creativity with the advance of technology to replace conventional creativity. Digital art tools like Photo Shop did not contribute to the creative process in a material way, it only facilitated the artist's existing ability. Modern AI models systematically steal and repurpose existing creative works while claiming that creativity as novel. Plagiarism does not facilitate creativity, it kills creativity.
Matthew McGuire
2025-07-27 17:45:40 +0000 UTCThe problem with AI isn't that it's better than artists, it's that it's good enough. That bland wall texture placeholder was there in the final game not because it was simply forgotten, but because it was good enough not to be noticed by anyone doing QA on the game. With AI image generation, we're moving towards an art industry where an artist is either a talented artists that sticks out above the sea of okay-ish AI generated art, or jobless. The Picasso's of the world will remain employed, but anyone studying the arts with the intent to make a living out of them better be extremely good or ready to fight the AI industry. I also think many people underestimate the capabilities AI has. What we're seeing now is the second or third version of a profitable AI product. It's the SNES of AI: popular enough to be taken up by the masses, but from an objective quality perspective extremely limited unless used to do unique and refreshing things by smart people. The concepts behind AI (by which I mean LLMs based on neural networks) go back to the 50s but were abandoned as theoretical science because all the computers in the world could not run the world's most basic neural net. Now, our phones are fast enough that they can extrapolate conversations from these vast networks of numbers. Commercial AI designs filter out swears, slurs, and other negative things not because AI cannot do them, but because ChatGPT/Google/Microsoft/Facebook doesn't want to risk the marketing damage done by an AI playfully insulting a user, but there are "uncensored" models out there that'll call you a motherfucking cunt if it deems that to be necessary. AI cannot do comedy *yet*, but you can clearly see the seeds in what it generates today, and given another decade or two of research, I'm sure it'll gain the ability to come up with witty jokes eventually. People take the developments in AI as if they just started five years ago, because that's when it hit the mainstream, but what we're seeing now is an explosion in AI research and capabilities. Seven years time have revolutionized a field of research that goes back at least seventy years, more if you count theoretical research. Seven years ago, GPTs were barely able to produce coherent sentences. Three years ago, they could do paragraphs but make up half of what they were saying. Today, GPTs are capable of turning a 2000 word prompt into 90% of an app. Comedians are safe for today, and probably for the next five years, but after that? AI video is getting good enough that it can take a script and output something that will get views on Tiktok/Youtube/Twitter/etc. and this is just the very first version; it's a matter of time before it can generate standup comedians and audience interaction. Plus, relatively well-known names like Yahtzee stand out above the crowd in a way that most budding comedians won't; making a name for oneself is becoming a lot harder with the noise that AI can and will generate. Going purely by technical capabilities, I fear that AI will be able to replace most employed artists in five years, maybe ten. Not because it matches its human counterpart, but because the people that pay for artists don't value the difference in quality enough to pay the difference between a subscription fee and a full living wage. We could band together and protest the use of generative AI in video games, but AI has already been normalized. I've seen quite a few writers enable tools like Grammarly to catch grammatical mistakes and such, but Grammarly is just ChatGPT telling you you've misspelt a word. Your phone's word suggestions are a miniature version of ChatGPT. Talking to Alexa and other "smart" home devices is talking into AI. Slowly these products weasel their ways into our daily lives, often without us noticing that there's "AI" behind them, until we come to see them as tools. Something not too dissimilar happened when tools like Photoshop first hit the scene; artists cried out that digital art is not real art, as the artist had the computer do all the work of aligning pixels and undoing mistakes. The artists that advocated for "pure" art over computer art lost out, big time, as the world digitized. Had they won out, I'm sure media of today would look extremely differently, with digital art being shunned unless it was a scan of a physical painting or drawing, a sentiment which likely would've killed the ZP style of animation back in the early days. With AI, we're approaching the same situation, for better or for worse. To support the many artists working in creative industries we should make our voices be heard, but I feel there's a certain irony in that the modern art industry feeding so many mouths today would've been a sliver of itself had the people that protested technological change in the art world fifty years ago gotten the support anti-AI artists get today.
Harige Knolraap
2025-07-25 08:55:10 +0000 UTCThe biggest ethical problem with Generative-AI is that it's built using stolen work without the creator's permission. It doesn't imagine, it appropriates. It's little more than a glorified plagiarism machine - a tool not for creating but for laundering content.
David C. Simon
2025-07-24 18:34:24 +0000 UTC
