Interactive fiction and generative AI

This post starts with a shoutout to Bruno Dias for three reasons:

1. For writing a blog post about generative AI that influenced a mammoth discussion thread on the Interactive Fiction Community Forums;
2. For having the ability to politely ignore a poster who was too clueless to realize that Dias not only wrote the post cited in their discussion, but had also actively commented in the thread; and
3. For the observation that this year’s AI-generated entries caught a rake in the face.

Maybe explicit rules aren't needed to ban generative AI from the Interactive Fiction Competition? Those types of rules are difficult to implement, especially for IFcomp, which relies on volunteers.

On the other hand, Parsercomp 2025 saw some drama related to generative AI and irregular voting patterns. There's a certain type of GenAI evangelist who persistently refuses to take a hint. Overall, I’m conflicted about the best way forward.

I mean, I prefer to treat judges like adults, which is an argument I've made regarding other concerns about “selection bias.” Start by assuming they can, you know, apply their own judgment, and have robust measures in place to prevent bad actors from gaming the voting system.

That’s what made the original post from Dias so alarming; it described a plausible scenario where rational judges had incentives to abandon IFcomp as a forum for creative human expression. 

After seeing how everything played out this year, I'm thinking that playing and rating entries (including at least a few GenAI works assessed fairly against their peers) is going to be an important part of ensuring that interactive fiction remains a safe space of creative weirdos.

(Elsewhere in game jams and creative weirdos, it's worth noting that November is the PicoSteveMo jam. In previous years, it was responsible for Dolan's Cadillac and Stand Elsewhere, so keep an eye on what develops this year.)

Image credit: lscottewart / Pixabay