Our managers want to use AI to generate performance evaluations and summarize Quarterly Conversations with their direct reports. How do we establish clear boundaries on where human empathy and candid feedback must remain completely untouched by AI?

Question

Tyler Smith · Accepted Answer

Using AI to evaluate human beings or script sensitive feedback is an abdication of leadership. The moment an employee realizes their quarterly review or compensation feedback was drafted by a machine, psychological safety vanishes and trust across your Accountability Chart is destroyed.

In the EOS model, the Quarterly Conversation is designed to build trust through open, honest, human dialogue around Core Values and GWC. AI has zero empathy, cannot read body language, and cannot perceive relational nuance.

Draw an absolute boundary in your company policy: AI may summarize quantitative performance facts, but human managers must own every word of evaluation and developmental feedback.

Specifically, managers can use AI to aggregate factual Scorecard numbers or check attendance logs across a ninety-day window. However, writing the actual assessment, evaluating alignment with Core Values, discussing whether someone has the GWC for their seat, and conducting the 5-5-5 conversation must remain one hundred percent human.

Make this distinction explicit in your team standards:
- AI analyzes operational output and logs raw milestones.
- Humans evaluate character, coach behavior, and deliver honest feedback.
- AI is strictly prohibited from drafting formal employee performance reviews or termination documentation.

Hold managers accountable in their Level 10 Meeting. If a leader lacks the capacity or willingness to deliver candid, empathetic human feedback without an automated crutch, that leader does not possess the GWC for a managerial seat.