I’m curious, how much are people using AI in your development process? Are you using it to debug chunks of code, are you using it to brainstorm plots or text or dialogue, are you explaining functions and asking AI to “vibe-code” your solution? Are you developing art because of your lack of art skills?
Are you literally attempting to incorporate an AI function or character within your game?
I don’t use AI when coding.
Just a heads up, there are plenty of people muting the AI tag right now, so your improvised poll may not be as representative as you might hope.
That said, no, I won’t be offloading any of my cognitive effort or friction, and thus personal growth, onto any LLMs or genAI systems.
Edited-to-add: Oh, and I know you’ve been preoccupied with life for a bit, so you wouldn’t necessarily know, but most of the Comps and Jams in this sphere either explicitly restrict the use of AI in submissions, or require full upfront disclosure upon entry. Between the two, if someone is working on something for a competition, it’s likely they wouldn’t use AI for that reason. Welcome back, by the way.
Why would I export the enjoyment of my hobbies to a machine that’ll just regurgitate someone else’s thoughts and work it stole? Isn’t the process part of the fun?
No, I don’t use AI and LLMs in any part of my process. I do not play games that use AI generated cover art, even as a placeholder, I do not play games that include any AI-generated code or text. If you can’t be bothered to take the time to make it yourself, why would I then spend my time on it?
All of my software, all except one package, was written well before AI was even a blip on the radar. I wrote my fractal software back in 2013-2016 while I worked for Statsoft which was later acquired by Dell. I got that job because I wrote a remote desktop system in 2000 which was downloaded a million times. I attribute most of my abilities to working there, at StatSoft. They were the best and the brightest. I just recently added an export function to save as a 3d file. I had all the data and researched STL files. They are amazingly simple when you already have a rectangular plane and elevation data. I’ve been meaning to do that work forever now but the need wasn’t really there because there are so many better fractal generators.
No I don’t use AI in any way.
I also worry that once you start using AI, you can come to rely on it, and use it more and more.
Pure artisanal code here.
I get that in this late-stage capitalism landscape people have to do objectionable things to make a living. I don’t fault anyone who has to use AI technology to keep their job. We’ve all got to live.
But as a hobby? Making toys? Nope. I don’t use it and I won’t have anything to do with projects that do.
I’m all for banning AI content from events.
But sometimes there’s an air of vitriol and self-satisfaction in here on this topic that I find unnecessary and frankly disappointing. I love everyone here, but I have to say it.
You’re probably right, Phil. I think everyone is both human and prone to a finite patience. Myself included. Something to keep in mind, though, so thanks for saying it
AI isn’t very good at Inform (and, I expect, choicescript), so I don’t use it there.
Since so many people use AI, I don’t use it for brainstorming since i feel like it just picks the most obvious choices, and for dialogue I like to try to create distinct voices.
There are a few places I’ve used AI:
-I’ve used context-sensitive online spellcheckers for a long time. They’ve performed better than offline spellcheckers for me. All the ones I know of incorporate AI right now. I don’t really care, since I just use it for typos, but it’s definitely an AI program.
-I’ve used AI to convert images to text, especially when it’s from a foreign language. Doesn’t come up in coding very often.
-I’ve used AI to suggest implementations of features in javascript that I am unfamiliar with. I then read the code to see if it makes sense. If it doesn’t make sense, I don’t use it. Even though, when I test it it often doesn’t work. I’ve done this to figure out how to force a hyperlink in a specific project (not bisquixe) to open a new window.
-I’ve used AI to read a specific chunk of C++ code and asked it to explain what was going on. It always gets parts blatantly wrong, but it has been a good start from time to time.
-I’ve asked AI to break down a foreign-language sentence or paragraph at a time that I hand wrote and explain my grammar errors to me.
Most of that isn’t useful to IF (except spell checking), so I tend not to use generative AI in my game programming. (The C++ stuff is unrelated, a different hobby project).
My problems with AI in interactive fiction are mostly performance-based. Most AI-written and AI-assisted games I’ve seen are bad. It’s possible that could be different one day. I do have vague fears of environmental damage in the back of my mind with it at times which makes me loathe to use it very much, but it’s not enough on its own given that I have a lot of similar concerns in a basket (like meat consumption and driving).
I am not too worried about concerns that it will replace artists or authors, since I think media will evolve around it (like it did with the introduction of photography or the way math evolved around calculators) and people will still be able to be creative and successful without it. But, for now, it sucks pretty bad at IF projects, and I haven’t seen any ways to both use it and have fun with a creative project at the same time (which is why I don’t use it for art).
[Edit: I mostly wanted to list it all like a confessional booth to avoid hypocrisy, as I am often vocally anti-AI in IF circles.]
I augment my existing programming and architecture expertise with GenAI. I would never use it for a story I claimed I wrote. I have had Claude write one story, Family Zoo, as a part of a tutorial for Sharpee. The story’s intent is to show elements of the platform. Not as any artistic output.
I’ve been going back and forth on how much I’d use Claude to generate code for a story I design and write and edging closer to very little, probably as a final sanity check on just the code.
I have a separate repo for my wips and the CLAUDE.md directives explicitly state where the line is: make no suggestion relating to the story or its puzzles.
I get the frustration from purists. I am obviously not one of them. I’ve already clearly stated my beliefs in several other threads and my Sharpee posts are AI tagged.
I do think having an AI do any writing is dumb. The storytelling in IF is the, to me, its creative foundation. You could look at the source of many 90’s and 2000’s games and they’d be far worse than how we measure vibe coding.
I enjoy writing and coding, so I have no desire to stop doing that. But I also think it’s not fair to ask my students to treat me as an authority on how AI works if I don’t have up-to-date experience with it, so I try to use it periodically for non-creative things entirely separate from my writing and code.
Recently I needed to take a list of thousands of nouns and figure out which ones were concrete (representing objects one can interact with) versus abstract (representing things that don’t have a physical existence). I have no qualms whatsoever about offloading that task to a neural network (especially one that I run locally to minimize the environmental impact), and that kind of text processing is exactly what large language models are best at.
Yeah, to be clear, I’m not fussed about local neural networks for non-creative tasks. But people usually ask about those specifically rather than about “AI” generally
I ask it for some CSS styles and some specific JavaScript code that I then refactor and modify myself.
I ask it how to do certain things in the painting program I’m using… or describe an art style technique that I want to read up on, but don’t know what it’s called (I just learned about rim lighting).
I ask it how I might make a sound effect softer in my sound editing program and maybe a strategy for composing chip tunes for the musically challenged.
I might ask it for a checklist (fill in the blanks) for thinking about a story theme or world building. Do I have a good foundation or have a missed a key ingredient?
I use AI quite often to get an answer or point me in the right direction (I used to Google that shit all the time before AI), but I refuse to have it generate art, audio, prose or code that I don’t maintain myself. I’m just not built that way. I’m interested in the process. I like the journey, even if it’s long, difficult and doesn’t even reach my intended destination.
I wouldn’t cry if AI was gone tomorrow.
That said, AI can be an invaluable tool. I just wish that people didn’t use it as a replacement for lack of ability or creativity. When something is built for human consumption, I want to see the craftmanship that was applied. I want to see a bit of the creator’s tastes and sensibilities at play. I want to connect somehow with the final result; and that extends to all aspects of that final result.
I don’t want to come off as an AI-defender, but in my experience, Claude (Opus) is surprisingly good at Inform 7, at least when it comes to analysing and fixing existing code.
Everything I make, I make myself. If I don’t know how to make it, I either learn or ask a friend for help. I don’t think I’m making anything complicated enough to have a machine do it for me, and even if I were doing some coding magic, I’d have no way to verify if it’s output is anything good, so I’d have to ask someone else anyway. When it comes to my prose, English isn’t my first language and I’ll probably keep making mistakes until I die, and I made my peace with it. If I’m worried, again, I ask a friend to read through.
Struggle is a part of the process. Some of the things others learned to recognize me by are a result of me adapting to artistic adversities along the way. If I had AI doing things for me, I wouldn’t develop certain methods of dealing with things which I enjoy. So, ethics and ecology aside, I just don’t use it because it’d probably make what I make awfully not-mine.
I use Claude Code (Opus & Fable) a lot as a coding assistant. I work with it on a daily basis as regional head for one of the departments of a global software company and I also use it to for my private projects. That being said, I never use is as a creative source and this is where I personally draw the line.
What it does: it buys me the time and the efficiency to maintain a project of scope and quality for the community such as The Puny BuildTools and most recently (actually not officially announced yet here) Arcturus, a programming language and compiler for the Infocom Z-machine.
I understand some of you are fundamentally against AI in any form and there are good reasons for that. Impact on the environment being one that bothers me the most personally. Unfortunately, history shows that you can hardly fight progress. We all ended up not living in caves anymore. We all ended buying a home computer in the 80s and we all have a smartphone in our pockets. Fun fact: as of today, more than 95% of the code in our company is implemented through AI under human guidance. It debugs faster and generates code that is more secure.
Being an author, artist and musician: anything that goes into the direction of creative replacement through AI: artworks, music, stories, games, that is something I can’t and will never support. For my personal projects: the tools I create and maintain for the community, such as the ones I mentioned above, are meant to make the process of writing and publishing wonderful human-made stories easier and more accessible.
There are many examples of “progress” that are now considered dead-ends, if they are remembered at all. There is no supersonic commercial flight anymore, and nobody is working on nuclear-powered cars.
It may well be that the current LLM business models turn out to be unsustainable, it ends up as an expensive niche product, and we all have to go back to programming the hard way.
Of course. I am explaining the state of the current reality on AI usage, so my comment is meant to be seen in context of that we currently deploy 95% of our code through AI (in our company). I did not intend to make a personal prediction on how AI will be successful or not and how the adoption is going to be in a few years from now. Betamax was also a brilliant video system in the 80s and everyone ended up using VHS instead. But since we touched now a possible future. Personally I don’t believe AI will go away. But I think it will develop into something more granular and local, moving away from the big power-hungry all-purpose models in the Cloud, at least partially.