Papers for

software designers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Ai workflows that match professional tasks improve analysis coverage and ease

"ChatGPT, what am I missing?": Designing AI Workflows around Professional Task Structure to Shape Analytic AI Use

Abstract: General-purpose AI lets users choose what support to request, but leaves them to structure the support a professional task requires. We examine how interactive workflows can embed professional task structure without prescribing how users engage with AI. We designed two scaffolded interfaces around the same negotiation scaffold: one presented a completed AI analysis, while the other supported user-directed, incremental development. A four-condition randomized experiment with 800 participants compared these interfaces with no-AI and an AI chat interface. AI-supported conditions improved preparation coverage over unaided work; the scaffolded workflows further improved coverage over chat. Although the scaffolded workflows produced similar coverage, the user-directed workflow elicited a broader repertoire of analytic requests and lower subjective effort. Professional scaffolding therefore depends not only on displayed structure but on how workflows organize users' engagement with it. Effective professional AI must structure how users and AI build analysis together.

Tue 15 SeptHuman-Computer Interaction
The gist
People using AI for professional tasks often have to decide how to use the AI’s help themselves. The authors studied two ways of organizing AI help: one that gave a full AI-written analysis and one that let users build the analysis step-by-step. Both helped users cover more important parts of their task than not using AI or just chatting with AI. Letting users direct the AI in building the analysis led to more varied questions and felt easier. How AI tools guide users to work together shapes how useful they are.
Open → 2609.16482v1

LLM assistants involving multiple people bring new risks and responses

When Single-User-Oriented LLM-based Assistants Involve Others: A Scoping Review of Pathways, Risks, and Responses

Abstract: LLM-based assistants are increasingly extending into multi-party contexts, while core operational processes for context management, personalization, identity attribution, authority attribution, and action execution often remain organized around a single user. Existing work examines particular multi-party settings, but lacks a systematic account of how these single-user-oriented assistants begin to involve additional human parties and what risks emerge. To address this gap, we conducted a scoping review of 58 studies. We identify five operational pathways spanning direct and indirect involvement, five recurring risk domains, and five areas of implemented and proposed responses. Based on these findings, we argue for governance that attends to changing cross-person roles and relationships in practice, and for assistant designs that preserve person-specific boundaries throughout interaction.

Sat 12 SeptHuman-Computer Interaction
The gist
Large language model (LLM) assistants are usually designed to help one person at a time, but they are increasingly used in situations with multiple people. The authors looked at 58 studies to understand how these assistants start including others, what risks come up, and how people try to manage them. They found five common ways others get involved, five main risk areas, and five ways people have responded to these risks. Their work suggests that assistants need better design and rules to handle interactions among multiple people while keeping each person's privacy and boundaries intact.
Open → 2609.14062v1