The Fourth Workshop on Intelligent and Interactive Writing Assistants

co-located with

The 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025)

Albuquerque, NM (in-person only)

The purpose of this interdisciplinary workshop is to facilitate discussion around writing assistants, thereby enhancing our understanding of their usage in writing process and predicting the consequences. To this end, we strive to bring together researchers from the human-computer interaction (HCI) and natural language processing (NLP) communities by alternating our workshop venue between HCI and NLP every year.

⬇️ Examples of new forms of human-machine collaborative writing. ⬇️

Important Dates

• Paper submission deadline: 28 February, 2025 11:59PM AoE (no extensions!)
• Paper acceptance notification: 12 March, 2025
• Workshop date: May 4, 2025

Sponsors

Tentative Schedule

• Keynote Speaker: Dr. N Katherine Hayles

  • 

  • Communicating with Aliens: Large Language Model as Writer/Critic/Reader

  • The arrival of Large Language Models (LLMs) has presented literary studies with unparalleled opportunities and challenges. LLMs communicate using human languages, an event never before accomplished; nevertheless, the ways in which LLMs achieve human-equivalent texts differ profoundly from how humans understand and employ language. Amidst the familiarity of machine-authored language lurks an alien sensibility based on mathematical abstractions, algorithmic processes, and deeply nonhuman embodiments. To grasp the significance of this alienness-in-familiarity, it is crucially important for users to understand what might be called the world-view, or umwelt, of the machine. Critics who argue that LLM texts have no meaning suffer from an anthropocentric bias that mistakenly insists the only meanings that count must be human-centric. This talk argues for a more productive approach that evaluates meaning-making within the context of the machine umwelt, comparing and contrasting it with human meaning productions.

• Speaker: Max Kreminsky

  • 

  • Interaction Techniques for Sculpting Expressive Text

  • Writing today goes hand in hand with typing. But what if writing was more like sculpting, or painting, or playing with toys? I discuss the wide range of potential writing modalities that might be enabled by natural language processing technologies – many of which allow texts to be squashed, stretched, split, merged, mutated, and warped in ways that bear little resemblance to traditional writing. New interaction techniques based on these new technological affordances, I argue, can complement more familiar processes of text production; promote the discovery of fortuitous but unlikely phrasings; and push expressive language out beyond its usual bounds. From the “new interfaces for textual expression” introduced by computational poet Allison Parrish to the playful creative writing instruments now under development in my lab, I explore a variety of strange computation-supported approaches to writerly interaction with expressive text.

• Speaker: Carly Schnitzler

  • 

  • Articulating Personal Voice with the Machine: Revision and Iteration with LLMs in Narrative Writing

  • I have conducted an IRB-approved study in which students share their work from the final personal narrative assignment in my Reintroduction to Writing: Digital Doppelganger course. In this assignment, students explore a uniquely human topic of their choosing and co-compose essays with LLMs over multiple iterative drafts, along with a revision reflection written without use of AI tools. The model for this assignment is Vauhini Vara’s essay “Ghosts.” Through the writing process, we examine how authorial power shifts with technology use, the utility and limitations of these tools, and the myriad ethical questions that arise with the use of AI in writing. This paper synthesizes findings from student reflections using a grounded theory methodology and advocates, as others have, that iteration and revision are central to students’ understandings of their own personal voices in writing. I argue LLMs can provide an opportunity to expand these writing processes and, when critically introduced and interrogated, can successfully intervene in a student-writer’s process of understanding their own personal voice.

• Speaker: Wei Xu

  • 

  • AI for All - Advancing Text Simplification and Reducing Cultural Bias in LLMs

  • The rapid growth of AI has increased information access, but linguistic complexity remains a barrier for many. This talk will focus on recent advances in Text Simplification, which aims to make written content more accessible while preserving its original meaning. Simplification can play a key role in broadening access to knowledge and countering misinformation, particularly for children and non-native speakers. I will present LENS :mag_right:, a learnable metric that improves large language models (LLMs) at the decoding time, and Thresh :ear_of_rice:, an interactive tool that helps humans evaluate LLM-generated text and trace errors back to the source. We also study cultural bias in LLMs, introducing CAMel :dromedary_camel: benchmark to assess Western favoritism in LLMs across multiple tasks and languages. Findings suggest a need for greater attention to implicit biases.

Accepted Papers

Organizers

You can contact the organizers by emailing in2writing.workshop@gmail.com.

Advisors

Call for Participation

Submit