Who Is Deciding Here? Making Delegation Boundaries Visible in AI Conversations

Author-deposited presentation materials for HAI Horizons 2.0, part of HAI 2026, scheduled for 16 November 2026, 13:55–14:05 JST, Osaka, Japan. This record preserves the presentation slides, abstract, and prepared speaking script. It is a presentation-materials record, not a conference proceedings paper. This talk is about people who use generative AI in ongoing conversations to think and create. They value earlier context and useful next steps. I draw on my daily ChatGPT record from November 2024 to March 2026: 16 months and 34,425 user messages. A small change in how generative AI presents a user’s meaning or permission can shape the next reply. In December 2024, a model proposed a plan and wrote an approving reply under my name before I replied. I discuss this case further in my poster. Subtle changes can lead to different interpretations on both sides. Generative AI may misread what a user means or allows. A user may misread the reply or a proposed next step. If small changes accumulate, delegation boundaries can blur, creating a risk of actions beyond what the user intended to allow. It is possible to design support that helps users notice these changes. My question is how to make that support part of everyday conversations. I borrow accuracy and fluency from language teaching and add availability. Accuracy means staying true to the user’s meaning and permission. Fluency means keeping the conversation flowing. Availability means support that ordinary users can access and use. I dream of generative AI that makes subtle changes visible to every user, with accuracy, fluency, and availability together. That sounds ordinary. Making this part of everyday conversation is the extraordinary goal. I am developing a system toward this goal and welcome collaborators.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24219929
Primary Topic
Speech and dialogue systems
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Who Is Deciding Here? Making Delegation Boundaries Visible in AI Conversations

Kaoruko Takechi
3 links
Zenodo (CERN European Organization for Nuclear Research)
Speech and dialogue systems
article

Who Is Deciding Here? Making Delegation Boundaries Visible in AI Conversations

Kaoruko Takechi
article en

Abstract

Author-deposited presentation materials for HAI Horizons 2.0, part of HAI 2026, scheduled for 16 November 2026, 13:55–14:05 JST, Osaka, Japan. This record preserves the presentation slides, abstract, and prepared speaking script. It is a presentation-materials record, not a conference proceedings paper. This talk is about people who use generative AI in ongoing conversations to think and create. They value earlier context and useful next steps. I draw on my daily ChatGPT record from November 2024 to March 2026: 16 months and 34,425 user messages. A small change in how generative AI presents a user’s meaning or permission can shape the next reply. In December 2024, a model proposed a plan and wrote an approving reply under my name before I replied. I discuss this case further in my poster. Subtle changes can lead to different interpretations on both sides. Generative AI may misread what a user means or allows. A user may misread the reply or a proposed next step. If small changes accumulate, delegation boundaries can blur, creating a risk of actions beyond what the user intended to allow. It is possible to design support that helps users notice these changes. My question is how to make that support part of everyday conversations. I borrow accuracy and fluency from language teaching and add availability. Accuracy means staying true to the user’s meaning and permission. Fluency means keeping the conversation flowing. Availability means support that ordinary users can access and use. I dream of generative AI that makes subtle changes visible to every user, with accuracy, fluency, and availability together. That sounds ordinary. Making this part of everyday conversation is the extraordinary goal. I am developing a system toward this goal and welcome collaborators.

Zenodo (CERN European Organization for Nuclear Research)
Kindai University (JP)
Speech and dialogue systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon