An Eye for AI: Eye-Tracking the Micro-Interruptions of GenAI Code SuggestionsArtifact Award Winner
Generative AI code suggestion tools, such as GitHub Copilot, are increasingly integrated into developer workflows. While these tools promise productivity gains, the actual impact on developer cognition and task performance has been mixed. In this paper, we present the first in-depth eye-tracking study of how developers interact with generative AI code suggestions during programming. We recruited 35 professionals and student developers and recorded their gaze behavior, code suggestion interactions, and code editing activity during programming sessions. By combining high-resolution eye-tracking data with fine-grained logging of AI-generated suggestions, we quantify the cognitive costs of reviewing code suggestions. Our findings show that approximately half of the generated suggestions are not even looked at. From the suggestions that were looked at, over 75% are not used. Although suggestions are only reviewed briefly (~0.9 seconds on average), each suggestion introduces a micro-interruption that disrupts developer flow. We discuss opportunities for more efficient and context-aware generative AI code suggestions that minimize cognitive overhead.
Thu 16 AprDisplayed time zone: Brasilia, Distrito Federal, Brazil change
16:00 - 17:30 | AI for Software Engineering 19Research Track at Oceania IX Chair(s): Fabio Palomba University of Salerno | ||
16:00 15mTalk | An Eye for AI: Eye-Tracking the Micro-Interruptions of GenAI Code SuggestionsArtifact Award Winner Research Track Pre-print Media Attached | ||
16:15 15mTalk | Inside Out: Uncovering How Comment Internalization Steers LLMs for Better or Worse Research Track Aaron Imani University of California, Irvine, Mohammad Moshirpour University of California, Irvine, Iftekhar Ahmed University of California at Irvine Pre-print Media Attached | ||
16:30 15mTalk | Scrub It Out! Erasing Sensitive Memorization in Code Language Models via Machine Unlearning Research Track Zhaoyang Chu Huazhong University of Science and Technology, Yao Wan Huazhong University of Science and Technology, Zhikun Zhang Zhejiang University, Di Wang King Abdullah University of Science and Technology, Zhou Yang University of Alberta, Alberta Machine Intelligence Institute , Hongyu Zhang Chongqing University, Pan Zhou Huazhong University of Science and Technology, Xuanhua Shi Huazhong University of Science and Technology, Hai Jin Huazhong University of Science and Technology, David Lo Singapore Management University Pre-print | ||
16:45 15mTalk | What Makes Code Generation Ethically Sourced?Distinguished Paper Award Research Track Zhuolin Xu Concordia University, Chenglin Li Concordia University, Qiushi Li Concordia University, Shin Hwei Tan Concordia University | ||
17:00 15mTalk | Filtering before Tuning: Robust Fine-Tuning of Large Code Models under Noisy Labels Research Track Zhong Li Nanjing University, Yang Chen China Automobile Data of Tianjin Co., Ltd. China Automotive Technology&Research Center Co.,Ltd., Heng Yong Nanjing University, Yuanyi Lin Huawei Technologies, Jiali Zhao Huawei, Tongtong Xu Huawei, Minxue Pan Nanjing University, Tian Zhang Nanjing University, Xuandong Li Nanjing University | ||
17:15 15mTalk | Automating Requirements Formalization: Using LLMs and Low-Complexity Distinguishing Traces for Semantic Validation Research Track Daniel Mendoza Stanford University, Anastasia Mavridou KBR / NASA Ames Research Center, Andreas Katis KBR / NASA Ames Research Center, Caroline Trippel Stanford University | ||