Paper Type
ERF
Abstract
Generative AI systems are increasingly embedded in knowledge work, yet their environmental consequences remain largely invisible at the point of interaction (Watson et al., 2010; Verdecchia et al., 2023), while productivity benefits are immediate and salient. This opacity creates a sustainability governance challenge: employees cannot factor environmental costs into usage decisions they cannot observe. We introduce two constructs to address this gap. Energy Disclosure Salience (EDS) captures the organizational practice of embedding ecological cost cues into LLM workflows; Perceived Ecological Impact Transparency (PEIT) captures employees’ beliefs that their organization communicates LLM environmental impacts in specific, credible, and actionable ways. Grounded in Signaling Theory and Social Learning Theory, we propose that EDS promotes pro-environmental LLM usage behavior (LLM-PEB), and PEIT independently drives LLM-PEB through self-regulatory mechanisms (Norton et al., 2015). The study employs a randomized experiment with knowledge workers interacting with a simulated LLM task environment. This research contributes to Green IS and sustainable AI research by integrating interface-level signaling mechanisms with perception-based interpretation processes.
Paper Number
1436
Recommended Citation
Nahin, Lutfi Al, "Green GenAI or, Dirty Infrastructure: Employees’ Perceived Ecological Impact Transparency of LLMs at Work" (2026). AMCIS 2026 Proceedings. 5.
https://aisel.aisnet.org/amcis2026/sig_green/sig_green/5
Green GenAI or, Dirty Infrastructure: Employees’ Perceived Ecological Impact Transparency of LLMs at Work
Generative AI systems are increasingly embedded in knowledge work, yet their environmental consequences remain largely invisible at the point of interaction (Watson et al., 2010; Verdecchia et al., 2023), while productivity benefits are immediate and salient. This opacity creates a sustainability governance challenge: employees cannot factor environmental costs into usage decisions they cannot observe. We introduce two constructs to address this gap. Energy Disclosure Salience (EDS) captures the organizational practice of embedding ecological cost cues into LLM workflows; Perceived Ecological Impact Transparency (PEIT) captures employees’ beliefs that their organization communicates LLM environmental impacts in specific, credible, and actionable ways. Grounded in Signaling Theory and Social Learning Theory, we propose that EDS promotes pro-environmental LLM usage behavior (LLM-PEB), and PEIT independently drives LLM-PEB through self-regulatory mechanisms (Norton et al., 2015). The study employs a randomized experiment with knowledge workers interacting with a simulated LLM task environment. This research contributes to Green IS and sustainable AI research by integrating interface-level signaling mechanisms with perception-based interpretation processes.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG GREEN