Paper Number
ICIS2026-3321
Paper Type
Completed
Abstract
Agentic coding systems now produce software faster than the workers accountable for it can read, evaluate, or trace. We theorize this shift through epistemic debt: the latent stock of work artifacts whose verification has been deferred by an accountable worker in a human-AI collaboration. We formulate the construct for AI-mediated knowledge work, distinguish three strategies — accept, negate, triage — for managing it, and examine it through 40 controlled Claude Code sessions, a repair experiment on 32 programs, a thematic analysis of 4,570 Reddit items, and six recorded sessions. We found that a community-derived constraint prompt produced a small, non-significant reduction in drift. The repair experiment showed when debt becomes consequential: repair burden rose when a fix required relational knowledge the code could not supply. The corpus reveals a 15-practice repertoire dominated by selective verification (92% triage). Tools and institutions designed for verification, not only production, are the next frontier.
Recommended Citation
Zhang, Shiyan and Nickerson, Jeffrey V., "When AI Codes Faster Than Workers Can Verify: Epistemic Debt in Vibe Coding" (2026). ICIS 2026 Proceedings. 22.
https://aisel.aisnet.org/icis2026/fow/fow/22
When AI Codes Faster Than Workers Can Verify: Epistemic Debt in Vibe Coding
Agentic coding systems now produce software faster than the workers accountable for it can read, evaluate, or trace. We theorize this shift through epistemic debt: the latent stock of work artifacts whose verification has been deferred by an accountable worker in a human-AI collaboration. We formulate the construct for AI-mediated knowledge work, distinguish three strategies — accept, negate, triage — for managing it, and examine it through 40 controlled Claude Code sessions, a repair experiment on 32 programs, a thematic analysis of 4,570 Reddit items, and six recorded sessions. We found that a community-derived constraint prompt produced a small, non-significant reduction in drift. The repair experiment showed when debt becomes consequential: repair burden rose when a fix required relational knowledge the code could not supply. The corpus reveals a 15-practice repertoire dominated by selective verification (92% triage). Tools and institutions designed for verification, not only production, are the next frontier.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
03_Future_of_work