Coaching Through AI Adoption Resistance

An adaptive conversation simulation designed to help managers practice navigating employee resistance to generative AI through listening, responsible-use boundaries, and practical experimentation.

Scenario Design · Adaptive Conversation Logic · Behavioral Coaching · Responsible AI · Change Enablement

Understanding the adoption challenge

Employees may resist AI for very different reasons: accuracy, confidentiality, increased verification work, professional judgment, leadership pressure, or uncertainty about how their role may change.

I designed this experience to give managers a safe place to practice uncovering those concerns before trying to solve them.

  1. Listen
  2. Understand
  3. Respond

Why practice matters

A manager can understand responsible AI principles conceptually and still struggle when an employee says, “I don’t trust this tool,” “I’m worried about our data,” or “What does this mean for my role?”

Rather than build another informational module, I designed a conversation where the learner has to respond in the moment and experience how different management choices affect the interaction.

Traditional learning

Explain the right approach

Experimental learning

Practice making the decision

Your conversation with Jordan

Jordan is a Senior Project Coordinator at Northstar Group who has worked with the team for seven years. Jordan is experienced, dependable, and cautious about generative AI after previously seeing an AI-generated client summary include information that was not in the source material.

You are Jordan’s people manager.

This is not a persuasion exercise. A successful conversation may lead to a small experiment, or it may lead to pausing until important questions are resolved.

About this experience: Jordan and Northstar Group are fictional. This AI-powered simulation was created for portfolio demonstration purposes. Conversations are processed in real time to power the experience and are not retained after the session. Please do not share confidential, proprietary, sensitive personal, or other private information during your session.

Jordan

Senior Project Coordinator

Before you begin

Choose Start conversation to practice by voice, or Start with text to type your responses. If you use voice, your browser will ask for microphone access.

Respond as Jordan’s people manager. Your goal is not to persuade him to use AI, but to understand what is behind his resistance, address legitimate concerns, and determine an appropriate next step.

There is no single “right” conversation. Jordan will respond differently based on how you approach him.

Coaching behaviors in practice

The experience uses the GROW coaching model as the conversation structure, with the Prosci ADKAR® Model as a lens for understanding resistance to change. Responsible AI principles are layered into the scenario to help managers move from listening and diagnosis to safe, relevant action.

  1. Listen & Diagnose

    Explore the current reality and understand the concern before advocating for a solution.

  2. Acknowledge Legitimate Concerns

    Treat accuracy, confidentiality, workload, professional judgment, and role concerns as meaningful information about the change.

  3. Establish Responsible AI Boundaries

    Clarify appropriate data use, verification, human judgment, and accountability before considering experimentation.

  4. Connect AI to Real Work

    Explore a relevant task or workflow rather than relying on broad claims about AI productivity.

  5. Agree on the Right Next Step

    Move toward an appropriate action when conditions are right, or pause when important questions remain unresolved.

Coaching structure: GROW Change lens: Prosci ADKAR® Application: Responsible AI adoption

How Jordan adapts

  1. Concerns are dismissed

    leads to

    Jordan becomes more skeptical.

  2. Trust is established

    leads to

    Jordan becomes more forthcoming.

  3. Concerns are meaningfully addressed

    leads to

    Jordan becomes willing to consider a limited experiment.

Decision gate

Not ready to experiment

becomes ready only when all five conditions are met

Five conditions

  1. Specific low-risk use case
  2. Clear data boundaries
  3. Review and verification process
  4. Human judgment and accountability
  5. Plan to evaluate the outcome
which leads to

Ready to experiment

The outcome depends on how the conversation unfolds, not on a predetermined response.

Responsible AI considerations

Data boundaries · Verification · Human judgment · Appropriate experimentation

Responsible AI considerations were incorporated into the scenario so adoption is not assumed to be the right outcome in every situation. The conversation encourages managers to consider data boundaries, verification, human judgment, and whether a proposed use case is appropriate for experimentation.

The persona also includes factual-consistency guardrails intended to limit unsupported assumptions about company policies, security controls, client details, or technical environments.

The public portfolio experience uses zero-retention settings so visitor conversations are not retained after the session.

Testing and refinement

Resistance — Passed

Trust — Passed

Low-risk experiment — Passed

Early testing surfaced several behaviors that needed refinement, including premature agreement, conditional assent before safeguards were established, unsupported persona details, and continued resistance after concerns had been meaningfully addressed.

I used those results to refine the persona rules, adaptive behavior, factual-consistency constraints, and decision logic, then retested the three core paths against the same prompt version.

From prototype to embedded experience

ElevenLabs Agent

Persona · Voice · Adaptive behavior · Decision logic

↓

React Experience

Voice + text · Transcript · Jordan identity · Controls

↓

Portfolio Embed

Responsive Squarespace experience

Built with: ElevenLabs Agents · React · Vite · Responsive HTML/CSS · Squarespace

Design choices: Fictional persona · Voice + text interaction · Hidden expression cues · Zero-retention public experience

What this work demonstrates.

Scenario Design · Adaptive Learning · AI Coaching · Manager Enablement · Responsible AI · Change Enablement

One takeaway from this project was the value of treating resistance as useful information. In this scenario, the most productive path was not to begin by convincing Jordan to adopt the tool, but to understand what was driving his hesitation, address legitimate risk, and determine whether there was a safe and relevant next step.