Composing with AI

Netprovs

One answer is netprov.

All these games and exercises with LLMs remind me that a good prompt, a few rules, and some guardrails can take writers of all skill levels far. Along those lines, my frequent collaborator Rob Wittig and I have developed three collaborative writing games on the topic of artificial intelligence, each with their own unique angle and context for play.

Netprovs are collaborative writing games that invite students and others to write together in response to prompts and a loose set of rules and guidelines. Wittig details these in his open access book. Netprovs are occasions for critical play, reflection, and “the voluntary healing of necessary relationships” (2022, p. 36). They are especially good at offering lighter responses to heavy anxieties.

Long before the release of ChatGPT and even Midjourney, we created a kind of summer school for artificial intelligence in The Machine Learning Breakfast Club. In this netprov, participants posted messages in an online teacher’s lounge for disgruntled educators tasked with helping recalcitrant AI.

Our second AI-inspired netprov was The Grand Exhibition of Prompts, which imagined a (not so) farflung future where AI has taken over the creation of all art, and the celebrated human artists exhibit their beautiful prompts. Modeled after the grand salons of Paris at the Fin de Siecle, our Grand Exhibition offered a chance of a lifetime for aspiring prompt artists from one of three schools: Retro (for the ironically nostalgic), Emo (for the deeply expressive), and Fido (for the pet-inspired). While developing their prompts, participant artists told the story of their (fictional) artistic lives.

During that netprov, another was born. In one of the discussion channels on the Discord server where we held our netprov, someone noticed autocorrect off by itself smoking a cigarette and looking sullen. Apparently, these anthropomorphized AI had pretty rough lives themselves. They could use some counselors of their own. That tangent gave rise to Pr0c3ss1ng, the support group for AI assistants. Allow me to quote from the call:

We are the secret human helpers who give artificial intelligence programs the courage to face the day. We’re the ones who hear Siri’s and Alexa’s tearful doubts and try to guide them in their stormy and complex rivalry. We’re the ones called on to help ChatGPT work through imposter syndrome. We’re the ones tasked to console Google Search as it sees all its parent company’s love going to the new Bard system. Not to mention the worries and resentments of legacy programs such as Autocorrect and Maps who feel eclipsed by the flashy newcomers.

Are we trained for this? No! Nobody is! We need support, too! That’s why we’ve created a discussion group where we can share stories, seek tips, and put our heads together to understand the intermachinal dynamics of the AI boom. You’re one of us! What do you see? What have you learned? Join us!

All three of these netprovs treat on the subject of AI as an occasion to reflect more deeply on what makes us human. Whether in performing the struggle of helping others or the ineffable qualities of artistic creation, players put their humanity on center stage, while AI served as a kind of fun house mirror or, at other times, a foil for us and our fragile operating systems.

AI can elicit quite a bit of fear, especially when it seems like it can perform our duties better than we can. I offer these activities as opportunities to do what humans still do best, to mess around, to test boundaries, not as service quality assurance testers but as unorthodox, unruly organisms who resist rules and rarely follow directions. To quote one of our players, when we presented him with the directions for our latest netprov, “Those are your rules.” He was determined to play by his own. Viva humanity.