From Future of Work to Future of Workers: Addressing Asymptomatic AI Harms for Dignified Human-AI Interaction
Citation. Ehsan, U., Passi, S., Saha, K., McNutt, T., Riedl, M. O., & Alcorn, S. (2026). From Future of Work to Future of Workers: Addressing Asymptomatic AI Harms for Dignified Human-AI Interaction. Proceedings of the CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3772318.3791081
Summary. This CHI 2026 paper argues that AI workplace adoption can produce an AI-as-Amplifier Paradox: short-run performance and efficiency gains can coexist with slower erosion of worker expertise, agency, and professional identity. Using a year-long study of AI deployment in radiation oncology with 42 participants, 24 interviews, 5 workshops, and 52 think-aloud sessions, the authors describe “intuition rust” and other asymptomatic harms that standard productivity dashboards may miss. They propose dignified Human-AI interaction and sociotechnical immunity as design frames for detecting, containing, and recovering from AI-induced skill erosion while preserving worker power and identity.
Key claims
- AI can be both enhancer and eroder: the same systems that improve speed, quality metrics, or throughput can reduce the practice, judgment, and identity through which expert work remains humanly and professionally meaningful.
- The radiation-oncology study found that AI-enabled efficiency gains were accompanied by subtle shifts in clinician behavior and self-understanding, including faster approval, reduced manual practice, fear of losing hands-on expertise, and concern about becoming “bystanders” in their own practice.
- The authors call these early shifts “asymptomatic effects” because they may not appear immediately in standard outcome metrics; over time, the effects can harden into chronic harms such as skill atrophy and identity commoditization.
- The paper defines dignified Human-AI interaction as AI integration that preserves human agency, expertise, and self-worth, rather than treating human-in-the-loop design as sufficient by itself.
- The proposed sociotechnical-immunity framework emphasizes early-warning signals, containment actions, and recovery routines that serve institutional quality goals while also giving workers more power to detect and respond to AI-induced erosion.
Evidence & limitations
- Evidence comes from a longitudinal field study in a high-stakes radiation-oncology setting, combining interviews, workshops, and think-aloud sessions; this is stronger evidence for worker-experience and expertise-erosion mechanisms than short-run productivity studies, but it remains concentrated in a specialized professional domain.
- The paper also evaluates the proposed framework across healthcare and AI-assisted software engineering, but the source is primarily qualitative and framework-building rather than causal estimation of productivity, skill decay, or patient/software outcomes.
- The arXiv version is public and lists the work as CHI 2026; the source should be read as a worker-centered HCI contribution, not as direct labor-demand evidence about AI substitution.