Our leadership team wants to use language models to draft employee performance reviews and score GWC quarterly conversations. What parts of our people operations and talent reviews must strictly stay human to avoid destroying team trust?

Question

Tyler Smith · Accepted Answer

Performance evaluations, Core Values assessments, and GWC (Gets it, Wants it, Capacity to do it) discussions must remain strictly human. The moment employees discover their manager used an AI prompt to write their review, personal trust evaporates and your company culture takes a direct hit.

You can use language models to compile raw operational metrics prior to a review. An automated script can pull an employee's Scorecard measurables, completed Rocks, attendance records, and project delivery deadlines over the trailing ninety days. Gathering objective data is low-value administrative work that software handles effectively.

However, interpreting that data and delivering feedback belongs entirely to the manager sitting in the seat. Evaluating whether someone truly 'Gets it' or shares your Core Values requires situational awareness, empathy, and personal accountability. A manager cannot outsource leadership courage to an algorithm.

If a team member is struggling, having a human conversation about root causes during your Quarterly Conversation builds respect and drives accountability. If an employee receives a machine-generated critique, they will become cynical and disengaged. Keep software focused on compiling objective numbers, and require your leadership team to own the words, the coaching, and the human relationship.