Back to the library

Key question New

AI-processed · Author-approved

How do employees keep their communication personal and distinctive when writing with generative AI?

Voice Work with GenAI: How Employees Develop and Sustain Individual Voice in Organizational Communication
Maayan Cohen, Lauren Waardenburg, Lior Zalmanson · International Conference on Information Systems (ICIS) · 2026
Illustration for: How do employees keep their communication personal and distinctive when writing with generative AI?

Loads Spotify's player; Spotify may set cookies. This device remembers your choice.

Key findings

  • Identified five recurring dimensions of 'voice work': protecting individual voice, articulating tacit preferences, externalizing voice into AI tools, refining an AI-supported mode of authorship, and rendering the production of voice visible.
  • Showed that employees frequently resist centralized, shared AI templates not out of tech resistance, but to preserve their unique relational styles with clients and colleagues.
  • Found that interacting with AI forced workers to make tacit communicative habits explicit, often causing them to inspect, evaluate, and deliberately alter how they communicate.
  • Demonstrated that workers maintain voice through continuous iterative calibration and negative rule-setting rather than one-time prompting or persona setup.
  • Revealed that prompt trails and interaction histories increasingly serve as visible artifacts for managers to evaluate employee judgment and problem-solving skills.

What it means for you

CIO / IT Executive

Decommission all centralized, mandatory AI text templates and instead provide employees with personal workspace storage to maintain their own iterative prompt libraries and negative rule-sets.

IT Manager

Audit your team's shared AI prompt history channels on Monday morning to evaluate how staff members exercise judgment and calibrate communication before deployment.

Business Strategist

Establish a team guideline that requires client-facing communications to incorporate distinct relational styles rather than generic AI outputs, protecting employee value and differentiation.

Researcher

Design a study examining how digital prompt trails and interaction histories function as proxy artifacts for managerial evaluation of employee cognitive labor.

Policymaker

Draft organizational guidelines that protect workers' rights to personalize AI-assisted communications, ensuring policies explicitly prohibit mandatory homogenization through uniform AI styling.

Summary

This qualitative field study examines how knowledge workers develop and preserve an individual voice while using generative AI for workplace communication. The researchers conducted a 13-month interpretive study at a 500-person advertising technology firm, combining non-participant observation, 56 semi-structured interviews, and digital artifact walkthroughs of employees' ChatGPT Enterprise workspaces.

Problem

Generative AI can draft competent workplace text, but its outputs often reflect standard model patterns that risk homogenizing organizational communication. In knowledge work, an employee's distinct communicative style signals judgment, builds relationships, and demonstrates ongoing human value, creating tension when generative tools write on a worker's behalf.

Transcript
Host: Welcome to A.I.S. Insights — powered by Living Knowledge. I'm Anna Ivy Summers.
Expert: And I'm Alex Ian Sutherland.
Host: Today, we are discussing a study titled "Voice Work with GenAI: How Employees Develop and Maintain Individual Voice in Organizational Communication." It was presented at the International Conference on Information Systems in 2026 by researchers Maayan Cohen, Lauren Waardenburg, and Lior Zalmanson. The study investigates how knowledge workers maintain a recognizable personal voice while using generative AI for everyday business writing.
Expert: That's right, Anna. Tools like ChatGPT can easily produce grammatically correct drafts, summaries, and client emails. However, because language models rely on statistical regularities, they tend to standardize tone and phrasing.
Host: Which creates a practical challenge for knowledge workers.
Expert: Exactly. In professional settings, how someone writes—their tone, vocabulary, and emphasis—signals their judgment, builds client relationships, and marks their distinct human contribution. When AI generates text on an employee's behalf, it introduces tension between automated efficiency and personal voice.
Host: How did the authors examine this issue?
Expert: They conducted a 13-month qualitative field study at an advertising technology firm with approximately 500 employees. Alongside observing meetings and training sessions, they conducted 56 interviews and carried out artifact walkthroughs, where employees shared their personal ChatGPT Enterprise workspaces, showing their prompts, project setups, and chat histories.
Host: What did they find about how employees manage their individual voice?
Expert: The researchers identified five dimensions of what they term "voice work." The first is protecting individual voice. For instance, a sales team avoided a centralized, shared email GPT built by their engineering department. They were not rejecting AI; rather, they wanted to preserve their unique relational styles with clients, which a generic template flattened.
Host: So rejecting a standardized tool was about preserving relationship nuances rather than resisting the technology itself.
Expert: Exactly. The second dimension is articulating voice. Human colleagues learn communication preferences tacitly over time, but AI requires explicit guidance. Workers had to state rules and examples explicitly, sometimes feeding past emails into the model to uncover their own subconscious writing patterns.
Host: And the third dimension deals with setting up the systems?
Expert: Yes, externalizing voice into AI arrangements. Employees created dedicated project spaces and custom instructions. Interestingly, some strictly separated work and personal queries, avoiding casual prompts like dinner recipes so they would not contaminate the professional profile the system maintained.
Host: That leads directly to the fourth dimension, refining authorship.
Expert: Right. Personalization was not a one-time setup; it required ongoing adjustment. Workers often refined outputs negatively—ruling out what was wrong. A recurring example was explicitly barring em dashes, because participants felt that punctuation mark had become a clear indicator of machine-generated text.
Host: And the final dimension is rendering the production of voice visible.
Expert: Yes. Traditionally, voice is judged only by the final message. With generative AI, the process itself leaves inspectable traces—prompts, revisions, and thread histories. In one case, a hiring manager asked job candidates to walk through their prompt threads to assess how they broke down problems and exercised communicative judgment.
Host: What are the main business takeaways from this study?
Expert: First, organizations should recognize that employee hesitation toward centralized, one-size-fits-all AI writing tools may not be resistance to AI, but an effort to safeguard client rapport and professional identity. Giving workers the latitude to personalize their own setups can support adoption. Second, as prompt trails become inspectable artifacts, managers have new ways to evaluate problem-solving processes, though companies will need to balance this visibility against employee concerns about monitoring.
Host: It highlights that adopting generative AI is an ongoing process of calibrating personal judgment alongside automated tools. Thank you, Alex.
Expert: My pleasure, Anna.
Host: And thank you for listening to A.I.S. Insights — powered by Living Knowledge. Join us next time as we examine another study on technology in the workplace.
Artificial Intelligence & Automation Information Systems & Management Human-Computer Interaction Data Analytics & Big Data Digital Transformation & Strategy Cybersecurity & Privacy Cloud & Infrastructure E-Commerce & Digital Platforms Operations & Supply Chain Sustainability & Green IS

generative AI, organizational communication, individual voice, knowledge work, human-AI interaction, qualitative research