As AI technology becomes more embedded in HR processes, it’s important for people leaders to maintain their focus on the human side of HR and ensure trust in the workplace.
By Cleo Valeroso
Artificial intelligence (AI) is now baked into nearly every corner of HR, reshaping how people hire, manage, pay, and interact with employees. But adoption alone doesn’t create trust. In fact, for many employees and candidates, AI has introduced more questions than answers.
Trust in HR processes isn’t something that can be automated; it’s not bundled in with every new application rollout. It’s earned over time with transparency, accountability, and humans firmly in the loop. When companies get that balance right, they build stronger engagement, better decisions, and more credible systems. When they don’t? The damage is real: demoralized teams, compliance risks, and a broken employer brand.
This is not a rejection of AI. It’s a call for responsible implementation, especially when real people’s lives and livelihoods are at stake.
AI in ATS: Efficiency but at What Expense?
AI-powered applicant tracking systems (ATS) have become the front line of recruiting. They promise faster screening by analyzing resumes against a checklist of keywords, formatting, and data patterns. But what they save in time, they often lose in context, creativity, and candidate experience.
As a result, job seekers are now tailoring their resumes for algorithms. That means plain formatting, buzzword-heavy language, and eliminating any flair that might confuse the parser. “Ugly but effective” resumes are now standard. Gone are the days of personalized design, creative storytelling, or even the inclusion of photos, which remain common in certain international markets. AI parsing strips away much of what once allowed a candidate’s personality to shine through.
This has eroded trust on the candidate side. A 2024 survey by HireVue found that while 73% of HR professionals are comfortable using AI in the hiring process, only 37% of job seekers share that confidence. That gap should concern every hiring leader, especially given the well documented consequences of AI bias in applicant tracking systems. Amazon scraped its AI recruiting tool after it started showing bias against women due to years of mostly male hiring data. Employers must use transparent systems to avoid similar failures. If the purpose, data, and limitations of an AI tool can’t be clearly explained, then it’s not ready for deployment.
Bias in AI: Not all Bias is Created Equal
Bias isn’t a bug in AI systems; it’s a mirror of the data they’re trained on. If that data reflects past hiring patterns, pay gaps, or performance ratings skewed by subjective criteria, then the AI will replicate and often amplify those biases.
But not all bias is inherently negative. In some cases, hiring managers may want to prioritize candidates who align with their team’s communication style or work culture. That kind of bias, if intentional, ethical, and transparent, can improve team cohesion and productivity. Still, this type of alignment should be approached with caution and documented logic. It can’t be an excuse for hiring by gut feelings masked by AI “objectivity.”
The problem isn’t just that AI reflects bias, it’s that it masks it under the false pretense of neutrality. That’s why regular audits are essential. McKinsey found that organizations with proactive AI audits experience fewer compliance issues and higher employee satisfaction. AI needs ongoing monitoring as practices and norms change.
HR Chatbots: Useful, but Only with Boundaries
I have developed an internal HR chatbot named Claire to help AI Squared access company verified answers to common questions, like time off, pay cycles, and policies, without waiting on an HR team member.
Used responsibly, HR chatbots can reduce administrative noise, help HR be a more strategic business partner, and empower employees to self-serve. But the company has built in tight controls: no access to open web search (to prevent hallucinated answers), no generative freeform replies, and limited access only to internally validated documents. Without those guardrails, a chatbot risks giving out bad advice or noncompliant guidance.
Transparency is key. Employees should be told exactly what the bot can and can’t do. If it pulls from company policy only, say so. If it doesn’t handle complex or sensitive requests, flag that clearly. Employees will only trust the tool if they understand its boundaries.
And every HR chatbot needs a backup plan: a human they can escalate to. Without a way to challenge it, a system without recourse is just deflection disguised as automation.
Compensation Benchmarking: Smarter When Contextualized
AI is now being used to benchmark compensation by aggregating salary data from external sources like Carta, Radford, and Levels.fyi. While useful, those tools only tell part of the story.
Public benchmarks often fail to reflect internal compensation philosophy, pay equity history, or role-specific nuances. A new hire in a high growth startup might require a different structure than someone in a mature organization with legacy pay bands.
The best approach blends AI generated market data with internal offer history, promotion timelines, and negotiated exceptions. This yields compensation recommendations that are fair, competitive, and aligned with the culture companies are trying to build, not just a market median.
If an employer is going to use AI in compensation planning, they also need to explain it. A candidate should be able to ask, “How was this offer calculated?” and get a clear, straightforward answer.
Humans Must Stay in the Loop, Always!
Whether evaluating performance, extending a job offer, or deciding on a layoff, these are not decisions that should be fully automated. The stakes are too high.
The European Union’s AI Act mandates human oversight for high-risk employment decisions, and so does SHRM. Even the best models can’t account for personal context, interpersonal nuance, or ethical gray areas. AI might flag an employee as an “attrition risk,” but it won’t know they’re dealing with a health crisis. It might suggest a promotion but not recognize the interpersonal conflicts a manager has been mediating.
The most successful AI implementations in HR are ones where humans make the final call, using AI as a smart assistant, not an unaccountable decision maker. As IBM puts it: “If we are to use AI to help make important decisions, it must be explainable.”
Automation can support fairness and consistency but judgment, empathy and accountability? That’s still an employers’ job.
In the Age of AI, Trust is the Real Talent Strategy
When implemented thoughtfully, AI can improve fairness, reduce administrative burden, and support better decision-making. But if employers lean too hard into automation without checks and balances and without human context, they risk losing the very thing that makes great workplaces work: trust.
Trust in HR isn’t about being likeable, it’s about driving results. When employees trust the systems and the people guiding them, they’re more engaged, more likely to share honest feedback, and more willing to invest in the company’s goals. That kind of trust turns policies into action, reduces resistance to change, and delivers better outcomes across the board, from retention, productivity, and innovation. In short, trust is the engine that powers real business results.
In HR, where the work is deeply human, technology should sharpen our judgment, not stand in for it. AI will only keep getting smarter. But if employers want systems people believe in, companies need to meet that progress with something only they can offer, clarity, context and care. That’s not resisting the future. That’s making sure everyone still belongs in it.
Cleo Valeroso is the vice president of global people operations and chief of staff at AI Squared.
