How to Use AI to Prepare for Difficult Conversations as a Leader

How to use AI for difficult conversations is less about finding the perfect script and more about preparing to lead well under pressure. Difficult conversations ask a great deal of leaders: clarity, empathy, steadiness, and sound judgment. AI can support that preparation by helping you think through your message, test how it might land, and practice your delivery without taking ownership away from you.

---

One of the hardest components of a leader’s role is delivering news that others don’t want to hear.

Avoiding these conversations comes at an enormous cost. Forty-three percent of employees waste two weeks or more ruminating about an unresolved problem at work, and one in three estimate their inability to speak up has cost their organization at least $25,000, according to a December 2021 study from Crucial Learning. 

At their worst, poorly managed difficult conversations create legal challenges. Research from Duke University found that “wrongful-termination claims were most strongly correlated with how workers felt they had been treated at the time of termination.” Multiply that by the number of employees in your organization, and you’re looking at a significant — and avoidable — expense. 

We know that words matter, as does the way in which you deliver them. However, the gap between knowing what to do and actually doing it is vast. 

How do you practice these encounters in a way that is safe and effective?

In the research for my latest book, AI for the Authentic Leader, I tested different AI tools and functionalities in order to understand which would most effectively address the challenges that keeps leaders up at night. Holding difficult conversations is certainly one of those use cases. 

While generative AI cannot hold these difficult conversations for you, AI excels at helping leaders prepare for those conversations. 

Using the voice mode of large language models like ChatGPT, Gemini, or Copilot, leaders can role-play those conversations, simulating different outcomes and preparing for the worst.  

Real-Life Example

In 2024, I faced a decision I’d been avoiding for months. After nearly ten years of building a global team, I decided to change the company structure. The business had grown in ways that no longer aligned with my vision. I was spending more time managing operations than doing what I did best—advising leaders on how to communicate amidst complex change.

I still remember the pit in my stomach after making the decision to downsize my team. I had to show up as my best self for each person. I couldn’t delegate these conversations, but AI could help me prepare.

For months, our company had used a custom chatbot which I built to serve as a Co-CEO to our business, based on an extensive prompt created by Paul Roetzer of SmarterX. I had customized the GPT with our company values, staffing, client personas, core services, and business goals, among other details. 

Here’s how my co-CEO chatbot helped me:

1. First, I typed my difficult messages into the chatbot and asked how well those messages reinforced the values of our company. 
2. Then, AI helped me fine-tune my messaging and the pacing of the conversation so that I could show the maximum level of respect and support for my team.
3. Next, I reviewed the AI suggestions, further adjusting the language so that it sounded like my voice and ensuring that it was authentic to me. I consulted a few trusted (human) mentors for extra guidance.
4. Finally, I used the chatbot’s voice mode to role-play the difficult conversations. I gave the AI non-private descriptions of the individual team members, such as their personality types and other sensitivities, and asked the AI to simulate different scenarios. The AI was endlessly patient. I would go for walks and talk through the conversations with AI as my sounding board. 

Ultimately, I was the one who delivered the message to each member of the team. It was painful every time. However, I was confident that I had done everything in my power to come across in the best possible way. While I was not perfect in how I handled them, I strongly believe those conversations went much better than they otherwise would have.

Addressing the Authenticity Question

How authentic was I when delivering those messages, given that level of practice and preparation? It’s a fair question, and one which I routinely receive when sharing this experience. 

In my work on authentic leadership, I define authenticity as, “speaking and acting in alignment with your values and beliefs.” This definition brings out who you are at your best, as opposed to you at your worst. Therefore, my use of AI enabled me to authentically represent the best version of myself. 

You might disagree. You might think I was too scripted and too unemotional when having those difficult conversations. But we’ve all been in situations where our emotions overwhelmed us and we said or did something we did not actually believe. 

Authenticity isn’t always a positive attribute: you can be authentically lazy, verbose, or abrasive. Those authentic traits can actively undermine your leadership effectiveness. 

Using AI helped me show up as a better leader. In fact, AI made me a better human.

Here’s what you can do.

• Identify a difficult message you plan to deliver. Perhaps you’re discussing changes in compensation structure or letting someone know they’re in the wrong career path.
• Brainstorm with AI. Input the message into an AI tool, along with your personal values and non-confidential details of the recipients of that message. Ask AI for different ways of relaying those message
• Use your own critical thinking. Evaluate the AI responses, using your own judgment to further refine the messages. Consult a human mentor or coach for guidance.
• Role play with AI. Ask AI to simulate different reactions, then provide feedback on how well you handle each situation. 
• Deliver the message. Ground yourself in your own values, remind yourself why this is important to you and to the recipients, and take ownership of your message.  

This example is not about outsourcing your authenticity. My goal is to ensure that the best part of you shows up when emotions might otherwise lead you to say something you’ll later regret.

Recall the statistics I shared earlier: employees wasting two weeks ruminating and organizations losing $25,000 per conversation. These costs are avoidable. 

AI won’t have the conversation for you, but it can help you prepare so you show up as the leader your team needs.

Until next week,

~Allison

---

A final thought as you prepare for these conversations:

One of the most powerful tools I teach is: pause and breathe. Before you respond, pause with your mouth closed and breathe through your nose.