Is there any interactive fiction showcase or competition left that will accept an AI generated cover or an LLM assisted parser? It seems while most commercial games these days flaunt their new AI technologies, this community wants nothing to do with it. Which is fine, but it leads me back to the question - is there any platform left to showcase Interactive fiction that has a little dab of AI-gen? If an open LLM were to replace the traditional parser, would any comp welcome that or will there be something new to rise up? Id like to imagine there will be some event in 2026 open to new AI technology.
Some discussion here: AI in competitions
You could look through itch’s jams to see if there are any: Game jams - itch.io
Or, of course, you could host your own jam/comp.
The IFComp rules for 2026 have not yet been announced. (Or debated, as far as I know – the surveys must still be trickling in.)
1 Like
Yeah, I think the Comp is the main one, though as Zarf indicates I wouldn’t be surprised if that changed for next year. So I do think folks who want platforms to showcase games and systems that use generative AI should think seriously about creating something built for the purpose.
(As a footnote, I don’t think most commercial games brag about their use of generative AI - as distinct from enemy AI and so on. My sense is that the corporate managers tend to tout the use of genAI on earnings calls because the promise of decreased labor costs is a positive for investors, but since the actual audience tends to dislike genAI, they typically stay mum or deflect when it comes time to actually market games)
9 Likes
Until recently, most of the comps allowed it. But the result was a lot of reviewers saying it killed their motivation to participate—and since a comp doesn’t work without reviewers (all the reviewer feedback is the best incentive and reward), the comps adapted by banning it.
The overall impression I get—more and more strongly over this past year—is that people enjoy generating games with LLMs a lot more than they enjoy playing LLM-generated games. So it’s going to be hard to find judges for an AI-powered competition, or even just a competition that allows AI in entries.
Which is unfortunate, because the technologies behind these LLMs do have enormous potential for text processing! While I have no interest in reading reams of LLM text, some of these components would be perfect for something like Aaron Reed’s smarter parser experiments (trying to gently guide newcomers to the proper format for parser commands by detecting common mistakes). A language model that can say “I understand what you’re trying to do, here’s how you would phrase that for a parser game to understand” would be a great way to make parser games more friendly to newcomers.
But the sheer amount of LLM slop being dumped onto the web nowadays, along with the ethical and environmental issues plaguing the current models, have rather poisoned the well for any future experiments. It used to be that Spring Thing was specifically the “anything goes” comp, but when reviewers said they would stop participating if it meant wading through increasing quantities of slop, the rules had to change—or else there wouldn’t be a comp to participate in!
9 Likes
5 posts were merged into an existing topic: AI in competitions
What happened to the “anything goes” initiative?
I don’t believe it ever got past the stage of being bandied about. Someone dedicated would have to host it, and most of the people discussing it were offering it as an alternative to comps that were clamping down on AI and, if I may editorialize a bit, offering it as a challenge to AI devs to see if anyone was motivated enough to run with the idea. Thus far nobody seems to have been.
2 Likes
That’s what i thought, thanks. I guess there’s an answer to OP’s question. Someone could carry the gauntlet of this idea. FWIW, i think the idea had additional possibilities outside of AI.
I’m still curious about the possibility of Penny Nichols-style games where ceding control to the LLM (for the sake of accommodating literally any player input) is the point, but I suspect that if there is a non-slop application for AI in IF then it’s going to be some kind of one-size-fits-all layer that translates invalid but sensible commands into something the parser understands. I imagine something much like @Draconis suggests, but with the inferred command actually carried out so newcomers don’t have to retype the “correct” version.
An AI-friendly comp might help arrive at or improve such a system, and early efforts might benefit significantly from the game itself being designed to make its task as easy as possible, but my feeling is that this sort of thing would ultimately make more sense as an extension of sorts than a game: the best possible version of it could be bolted onto just about anything, and wouldn’t necessarily make sense as something to be judged as a competition entry.
3 Likes
This is true, but also, most people are drawn to UI things when they see a game using them and go “hey this is really slick”. I pivoted to all the hyperlinks in my recent parser games after playing The Impossible Bottle.
2 Likes
9 posts were merged into an existing topic: AI in competitions
A post was merged into an existing topic: AI in competitions
I’ve moved some general AI discussion to an existing topic. Please reserve this topic specifically for existing or upcoming competitions and jams that solicit or permit AI as per the title: Interactive fiction competitions open to genAI entries - 2026
If you’d like to speculatively discuss how AI works generally in regards to competition entries, please participate in this topic:
7 Likes