Cara Menulis Penilaian Kendiri atau Semakan Prestasi dengan AI

The most useful way to think about AI here is as an organizer and editor, not a source of facts. A model knows nothing about your year except what you paste in, so the work starts with raw material: a running 'brag document' if you kept one, your goals from the last cycle, project summaries, metrics, thank-you messages and feedback you received. Next, give the AI your company's review criteria or competencies and ask it to sort your evidence under each one. A common structure for each accomplishment is situation, action and result: what the problem was, what you specifically did, and what changed. A productive technique is to have the AI interview you. Ask it to question you one item at a time about each project, including what you personally contributed and how you know it worked. This brings up details you would otherwise forget. Then have it draft, and edit the draft so it sounds like you and every claim is true. Two misconceptions cause most of the problems. First, asking the AI to 'make this sound impressive' invites invented numbers and inflated language, and a reviewer who knows the work will notice. Second, people assume a self-review should list only wins. Many managers respond better to an honest growth area paired with a plan. For managers, privacy is the central caution. Reviews contain personal information about other people. Depending on account settings, consumer chat tools may keep conversations or use them to improve their models. Follow your employer's AI policy, prefer approved enterprise tools, and replace names with roles or initials. Managers should also watch for bias. Research on written reviews has found that women more often receive vague or personality-focused comments than specific feedback about their work. Ask the AI to flag feedback that describes personality rather than behavior, but the judgment about the person remains yours.