Jonathan Cheng

@jonathancheng.bsky.social

English Literature PhD turned ML Researcher. I convert cold brews into noisy experiments. Foundation Models @ Apple NYC 🏳️‍🌈

Which all just reeks of what happens when a status order gets upended by some new technique. Basically everyone who is under threat runs around shouting “Never mind what I appeared to value before, what I *really* value is this incredibly intangible thing I possess that also preserves my status”.

New from our lab! #COLM2026 When people generate stories, they don't just write one prompt. Instead, they explore narrative space via branching edits 🌱 We reconstruct 24k of these edit trees 🌳 from chat logs and map edit types, story formats, how they relate to tree depth, and more!

The Garden of Forking Prompts: How Users Explore Narrative
Space in Story Generation
Advait Deshmukh♣ Nora Benedict♠ Melanie Walsh♡ Maria Antoniak♣
♣University of Colorado Boulder ♠University of Georgia ♡University of Washington

Abstract

Large language models (LLMs) have changed the way people engage
with stories. Drawing on public chatbot logs, we can see that when users
generate stories, they iteratively edit their prompts to explore narrative
possibilities, adjusting characters, redirecting plots, and swapping fictional universes. As aggregated data, these prompts represent rich traces of creative preference at scale. Yet story generation evaluation benchmarks rely on static, one-shot prompts that cannot capture this exploratory behavior. In this work, we study how users revise consecutive story prompts in the wild. Using a dataset of naturally occurring user-chatbot conversations, we construct WildStories, a sample of 275,635 story generation prompts (labeled with story format, prompt components, and explicitness), and WildEdits, a collection of 24,291 edit trees that model how users iteratively edit base story prompts and explore branching story possibilities. From these trees we develop a framework of edit types crossing four directions (adding, removing, changing, and extending) with fourteen targets (e.g., plot, character, genre). We then use our datasets and this framework to
analyze user behavior in navigating narrative space via LLMs. Finally, we
show how automated permutations based on the framework can be used
for story generation benchmarking. Content Warning: This paper works with “wild” chatbot logs, which often include toxic and sexually explicit themes.
Advait Deshmukh@advaitdeshmukh.com · 3w ago

1/7 In 1941, Borges imagined an impossible novel that follows all narrative branches at once. Today, we can observe chatbot users exploring narrative possibilities as they repeatedly edit story prompts. In our COLM 2026 paper, we studied this branching exploration via the WildChat dataset. 🧵

Paraphrased edit trees from the dataset

Someone asked me what my ideal work environment looks like. And all I can imagine is me swiveling my chair, with a valley girl accent, saying “you already tried to increase the batch size, don’t do it, girlfriend!🌸”

TIL that my dad, who wasn’t close with biological parents, was wrong about his genealogy. He always thought he was 1/2 Spanish, 1/2 Chinese. TIL I learned that I’m 1/4 Peruvian! And I have a bunch of cousins in Peru and *Wales* 😂

Got around to reading Piransei. Anyone who saw me on the subway, would’ve seen me *really* trying to get into it. My take: The Book is a Hall of great Beauty in which very little happens. I walked its length twice and found the same Statues, and the Tide, when it came, did not carry me anywhere.

About to pose the classic weeb dilemma to my partner, would you rather watch: * a deeply serious and beautifully animated film about an aging Japanese actress * or… an anime about an idiosyncratic teenager using his power to usurp an American empire — also, robot fights!

if you think students will be routinely collaborating with LLMs after graduation, #1 priority is to study rhetoric because you need to recognize BS and critique the rigor of an argument that *looks* good. the elite users are going to be like, philosophy, classics, and english majors.

Just watched Six the musical, and I’ve discovered: 1) the closer a musical is to a concert, the more I prefer it 2) Kirstin Maldonado from Pentatonix is one of the cast members right now!!! I think my favorite musical experience I’ve ever had?

The confluence of surviving gargantuan deadlines, catching up on research I wanted to do, and bingeing the new Olivia Rodrigo album is really, really hitting the right way. Oh, what’s that? Sun is shining, iced coffee in hand, and all my training jobs are emitting healthy logs? *skips*