When AI summarizes the performance data, why managers feel even more lost

In an era when AI summarizes performance data, why have managers gotten busier

After AI entered performance management systems, many managers find themselves in a similar situation. A report in which AI has summarized a team member's annual performance activity arrives by email. There is a 1:1 meeting tomorrow. Open the report and the check-in completion rate, feedback keywords, and goal attainment rate are all neatly organized. Yet, sitting in front of this data, the manager thinks: "The data is clearer than ever, but I have no idea what to actually do with it." Even though AI now summarizes annual performance activity, analyzes feedback text, and auto-generates evaluation reports, there is a reason managers feel busier than before.

This scene is not a one-off personal experience. In a Gartner (2026) survey of 2,986 employees, 46% of managers were already using AI in their work, while 86% of them reported experiencing difficulties of varying degrees in leading their team's effective use of AI. The tools exist, but the know-how does not.

The problem is not the AI tool itself. It is that the manager's role is unclear in the process of turning AI-summarized data into actual conversations with team members. The gap between the expectation that "AI does it all" and the reality that "in the end, I have to have the conversation" is creating confusion.

What AI reduces, and what it cannot reduce

According to a joint study by Harvard Business School and the global consulting group BCG, managers who adopted AI in their work cut the time spent on management administrative tasks—such as data analysis and document writing—by an average of 25%. That is a significant figure. Yet few managers have been clearly guided on what this freed-up time should be filled with.

What AI reduces is clear: the time spent reading and summarizing annual performance records, the time spent finding patterns in feedback text, and the time spent drafting evaluations are prime examples. On the other hand, the areas AI cannot replace are equally clear.

Misreading this distinction leads to two kinds of failure. One is the misconception that "AI has summarized it, so the conversation can be skipped"; the other is the illusion that "we have the AI data, so just showing this is enough." From the team member's perspective, both read as a signal that the manager has no interest in them.

In fact, "a decline in employees' trust regarding fairness and transparency" was cited as one of the biggest risks when adopting AI performance management (Gartner, 2024). When a manager presents an AI-generated summary or evaluation as-is, the team member feels, "they see me as data, not as a person." This is why it is essential for the manager to digest what AI has organized and deliver it in their own words.

What comes after receiving the data is the manager's real work

If we think concretely about what a manager should do after AI summarizes the performance data, it ultimately converges on three things.

First, turning data into questions. If AI summarizes that "this team member had a 60% check-in completion rate in the first half, and negative expressions related to collaboration recurred in peer feedback," the manager's role is not to relay this fact as-is. It is to turn it into a question: "Check-ins didn't go so well in the first half—was there a point in your workflow where you felt stuck?"

Second, filling in the context the data fails to capture. AI only sees recorded data. The circumstances under which a team member achieved that result, and what their invisible contributions were, are things the manager has observed and remembered firsthand. Without this context, data is nothing more than numbers.

Third, connecting the data to the starting point of the next stage of growth. If the AI summary is "an organization of past performance," the manager's role is to design "the goals and development direction for the next period" together with the team member. Without this connection, a performance review ends as a report on things already past.

In this way, performance management in the AI era paradoxically leaves only the "most human domain." Administrative aggregation and summarization are handled by AI, while deep understanding of and conversation with people concentrate entirely on the manager. With the arrival of data, a manager's work becomes less mechanical and, at the same time, evolves to be far more human-centered.

The organization that turns AI-summarized data into conversation

Turning this flow into an actual operating structure is the core task of adopting AI performance management. The system that surfaces the data and the manager's role in turning that data into conversation must be designed together.

Maintaining this structure by hand is realistically difficult. elizax's AI performance data summaries and AI evaluation summary reports provide a structure that reallocates the time a manager used to spend collecting and organizing data so they can focus on conversations with team members. talenx's 1:1 meeting feature is designed so that AI-organized performance data and talent snapshots can be viewed together within the context of a 1:1 conversation.

What matters is the order. A structure in which AI first organizes the data, and the manager focuses on people on top of it. Reverse this order—if managers lose time to talk with people because they are busy organizing data—and the benefits of adopting AI come out at less than half.

AI summarizing performance data does not mean a manager's role shrinks. It means managers can spend more time on what they truly should do—understanding their team members, having conversations, and designing growth.