For HR Managers and People Partners
Cognitive Sovereignty Checklist for HR Managers
AI tools in recruitment screening, employee relations, and HR analytics can invisibly narrow your thinking before you ever engage with a person. You may find yourself trusting an algorithm's recommendation instead of your own judgement about fit, potential, or performance. This checklist helps you stay in control of the decisions that shape people's careers and your organisation's culture.
These are suggestions. Take what fits, leave the rest.
0 / 19 applicable
Tap once to check, again to mark N/A, again to reset.
Recruitment: Keeping Your Judgement First
Review the full candidate list before AI ranking appearsbeginner
Before opening your LinkedIn Recruiters AI results or HireVue screening report, ask your team to pull the unranked candidate pool. See who applied. Your first impression matters and shapes what you notice in AI recommendations. AI ranking can anchor your thinking toward its preferred candidates.
Name the specific trait you are assessing before reading the AI summarybeginner
Before you look at what ChatGPT or an AI screening tool says about a candidate, write down what you actually want to assess in this role. Are you looking for problem-solving ability or domain knowledge or communication skill? This stops you from accepting AI's framing of what matters.
Always read the original application or CV after the AI assessmentbeginner
Do not skip the source material. Read what the candidate actually wrote. You may notice something the AI flagged as a red flag that you interpret differently. You may spot potential the AI missed entirely. Your reading of someone's words is a separate source of truth.
Ask your AI tool to explain why it rejected a candidateintermediate
Workday AI and HireVue often filter candidates silently. Request the reasons for rejection. If a tool rejects someone because they had a career gap or changed industries, you need to know that before you agree with the decision. Some of your best hires may have non-standard paths.
Track which candidates you would have rejected versus which your AI tool rejectedintermediate
Keep a simple record for one hiring cycle. Note candidates the AI rejected that you would have shortlisted anyway. After three months, check whether those candidates performed better or worse than AI-recommended hires. You need data about where your judgement differs from the algorithm.
Conduct at least one interview stage without seeing the AI scoreintermediate
Have a colleague conduct a first-round phone screen without access to the AI assessment. Compare notes afterward. This shows you whether the AI is influencing how you interpret what a candidate says in real time.
Identify which job criteria your AI tool cannot assessadvanced
LinkedIn Recruiters AI and similar tools screen for keywords and experience patterns. They cannot assess cultural fit in your specific team, appetite for learning, or how someone handles ambiguity under pressure. These are the areas where your judgement must be the primary filter.
Employee Relations: Protecting Conversation Quality
Draft sensitive communications yourself before using AI as a second draftbeginner
When handling performance concerns, disciplinary matters, or resignation responses, write your own first version. Then use ChatGPT or similar tools to check your tone. Reversing the order means the AI template shapes what you say instead of your understanding of the specific person and situation.
Identify which employee conversations should never be AI-mediatedbeginner
Feedback on underperformance, discussions about mental health struggles, or conversations about discrimination or harassment need your voice and presence. If you are using Workday to send a standard template response, you have outsourced something that requires your judgement.
Notice when you are using AI to avoid difficult conversationsintermediate
If you find yourself drafting a tough message with AI, pause. Are you using the tool because it genuinely improves communication or because it feels safer to hide behind polished language? Sometimes your direct, honest words are what an employee actually needs to hear.
Read employee feedback and survey responses yourself before reading AI summariesintermediate
HR analytics tools often cluster feedback into sentiment scores or themes. Read the actual comments first. Employees often express contradictions and nuance that an algorithm flattens into a category. Your reading of what people said teaches you something different than a summary.
Ask your analytics tool which comments it excluded or weighted as outliersadvanced
AI-driven HR analytics often identify 'consensus' views. The outlier voices may be the employees struggling most or the ones with the most important insight. Request the full list of comments flagged as statistical outliers so you can judge their importance yourself.
Conduct one-to-one check-ins without reviewing the employee's digital performance score firstadvanced
Rippling AI and Workday generate performance indicators from email activity, meeting attendance, and project completion rates. Go into a conversation with someone without that score. Talk to them first. Then compare what they told you with what the data said. Notice the differences.
Performance and Policy: Staying in the Decision-Making Role
Write your performance management policy without starting from an AI templatebeginner
Your organisation's approach to performance management should reflect your actual values and how you want managers to lead. Using a ChatGPT policy as your first draft means you are building your culture from generic best practice instead of your own thinking. Draft what matters to you first.
Audit a sample of performance ratings for patterns that favour AI-screened candidatesintermediate
Employees hired through AI recruitment screening may be rated higher simply because they started with algorithmic approval. Review ratings for a cohort hired by AI versus a cohort hired through traditional screening. Look for rating inflation among the AI-selected group.
Notice when you are accepting an AI recommendation without asking whyintermediate
Every time you act on a tool's suggestion, pause and write down why you agreed. Was it because you actually think it is right or because the AI said so with confidence? After two weeks, review your notes. You will recognise patterns in when you stop thinking.
Create a policy for when your judgement overrides an AI recommendationintermediate
Do not leave overrides informal. Document circumstances where you rejected an AI recommendation and explain why. This creates a record and forces you to articulate your reasoning instead of silently accepting the algorithm's choice.
Test your HR analytics tool on a decision you already know the answer toadvanced
Ask your analytics platform to identify high-flight-risk employees or top performers. Compare the results to your own knowledge of who is actually engaged or likely to leave. If the tool misses obvious cases, you know where it is unreliable in your context.
Require managers to justify departures from AI-generated performance improvement plansadvanced
If your system generates standard performance improvement plans, tell managers they must explain if they modify it. This protects individual circumstances from being flattened into a template and keeps your managers accountable for their own judgement.
Five things worth remembering
- Before you use an AI tool for any HR decision, ask yourself whether you trust a system more than you trust yourself. If the answer is yes, stop and rebuild your confidence in your own thinking before proceeding.
- Keep a three-month record of one decision type where you use AI (hiring, performance rating, or policy). Compare your AI-guided choices against your non-AI choices six months later. Let the results teach you where algorithms help and where they mislead.
- Create a standing agenda item in your team meetings to discuss where you disagreed with an AI recommendation and why you overrode it. This builds collective confidence in human judgement and stops the tool from becoming unquestionable.
- Never let an AI tool be the first filter for anything. You move second. You review what it catches. You decide what it missed. Reversing this order means the algorithm shapes what you see instead of you shaping what matters.
- Talk to your frontline managers about their experience with AI in hiring and performance management. They see whether AI recommendations actually predict success or whether they just match a profile. Their feedback is more valuable than any accuracy metric the vendor provides.
Related reads
Common questions
Should hr managers review the full candidate list before ai ranking appears?
Before opening your LinkedIn Recruiters AI results or HireVue screening report, ask your team to pull the unranked candidate pool. See who applied. Your first impression matters and shapes what you notice in AI recommendations. AI ranking can anchor your thinking toward its preferred candidates.
Should hr managers name the specific trait you are assessing before reading the ai summary?
Before you look at what ChatGPT or an AI screening tool says about a candidate, write down what you actually want to assess in this role. Are you looking for problem-solving ability or domain knowledge or communication skill? This stops you from accepting AI's framing of what matters.
Should hr managers always read the original application or cv after the ai assessment?
Do not skip the source material. Read what the candidate actually wrote. You may notice something the AI flagged as a red flag that you interpret differently. You may spot potential the AI missed entirely. Your reading of someone's words is a separate source of truth.