How do you provide creative freedom without getting things, without having system bias s or other stuff creep in? Like, if you're pulling in all these texts, do you inadvertently wind up with texts that are misognistic and biased? And how much do you have to vary very deliberately say, we've got to make sure weare looking out for these things that ai tends to get sucked into. That's a wonderful question, because it is a very real danger of building anything with large language models which are largely trained off english on the inner nut.
Sure. GPT-3 and large language models in general can take a prompt and spin out in any of a million human-sounding directions. That's neat, but maybe not exactly what you'd want to turn loose as your guide through a narrative multiverse of AI-boosted creative play and community. "A what?" You say. Exactly. In this episode, we dug into Hilary Mason's latest endeavor, Hidden Door, and how she and her team are working to apply the right level of "human" to AI-driven narrative play. Intrigued? You should be! It's fascinating! For complete show notes, including links to items mentioned in this episode and a transcript of the show, visit the show page.