AI is becoming HR's second opinion on pay and promotion calls, and it is overturning managers' decisions. The harder question is who checks the second opinion, just as merit and promotion season arrives.
Key Takeaways
For years the worry about AI in people decisions has been that it would bake bias in. A new survey points the other way, at least on the evidence HR teams are reporting. HiBob's research, released November 12, found that 57% of HR professionals using AI to review workforce decisions have reconsidered an employee who was initially passed over for promotion. Nearly three in four have taken AI information to a manager and asked them to think again.
That is a meaningful shift in who holds the pen on pay and progression. It also lands at the moment most organizations are heading into year-end merit and promotion cycles, when the volume of these decisions peaks and the time to scrutinize each one is shortest.
The HiBob data shows how far AI has already moved into consequential calls. Respondents said AI informs performance ratings (64%), promotions (51%), pay rises (50%) and redundancies (42%). The most common uses are analyzing workforce information (47%), screening candidates or reviewing hiring recommendations (45%) and preparing or reviewing performance assessments (40%). Almost a third (31%) use it to plan changes to jobs, skills or how work is organized.
The outcomes respondents describe favor employees. 43% report a more favorable final decision on pay rises after an AI review, 42% on performance ratings, and 40% on promotions. Of the 73% who used AI information to push a manager to reconsider, 42% did so because an employee's or candidate's skills or potential had been overlooked. "AI is fundamentally changing how leaders assess capability and build teams," said Ronni Zehavi, HiBob's CEO and co-founder, who framed the tool as a second opinion that lets HR test assumptions. HR Dive's weekly review led with the obvious question the numbers raise: can AI root out manager bias?
Two cautions belong next to those figures. HiBob sells HR software, and its release does not disclose a sample size, fieldwork dates or where respondents are based beyond a London dateline. And the results are what HR professionals say happened, not measured changes in who got promoted. Treat them as a signal of how HR is starting to use the tools, not as proof that AI makes decisions fairer.
The reason HR is reaching for a second opinion is that the first one has a credibility problem. A HiBob survey of 4,700 people managers reported by HRreview in February, conducted by Censuswide between late December 2025 and early January 2026, found that 83% believe they can clearly justify salary and promotion outcomes with data. Yet more than three-quarters had a formal appeal against a pay or promotion decision in the past year, including 76% who faced challenges over promotion allocations.
The same managers admitted the data underneath their decisions is thin. 30% said a lack of timely data played a large or very large role in a high performer being underpaid or under-recognized, and more than a quarter said it contributed to promotions awarded before people were ready. More than two-thirds worried that similar roles were assessed using different criteria across the business, nearly two-thirds said they had made an educated guess rather than miss a deadline, and only 2% had a single dashboard combining HR and finance data. "Unifying that data doesn't just make decisions faster; it makes them fairer," said Toby Hough, HiBob's vice president of people and culture for EMEA.
Employees have been saying something similar for a while. In a 2023 Syndio survey of more than 1,000 workers, about one in four said a supervisor's personal biases had hurt their performance review, 43% of non-management employees said their organization lacks transparency in promotion decisions, and two-thirds said they felt uncomfortable expressing interest in advancement for fear of bias. Overlooked talent is not a new discovery. What is new is HR having a tool that surfaces it at the moment a decision is being made.
The gap in the HiBob data is in the safeguards. 59% of HR professionals say AI information is checked for accuracy and relevance, but only 45% review proposed decisions for unfair or inconsistent treatment, and 41% consider individual circumstances the AI may have missed. Put differently, a majority of teams using AI on promotions are not systematically checking whether the AI-informed outcome is consistent across people.
Adoption is also uneven. 38% of respondents do not use AI information for pay-rise recommendations, 33% do not use it for promotions, and 39% keep it out of grievances and disciplinary matters. That restraint is reasonable, but it means two managers in the same company can reach the same kind of decision with and without an algorithmic check. If AI helps one group of employees get reconsidered and not another, the tool meant to reduce inconsistency becomes a new source of it.
There is also a manager problem the technology cannot solve. An AI flag only changes an outcome if a manager accepts the challenge, and if the reasoning can be explained to the employee afterward. Teams that treat the AI as a verdict, rather than as evidence a manager has to weigh and defend, risk replacing one unexplained decision with another.
AI catching a passed-over employee is a good outcome. It is also an admission that the original process missed them. HR's job this cycle is to make sure the second opinion is applied consistently, checked for its own blind spots, and owned by a manager who can explain the final decision.
Checklist
The steps and data to line up before merit decisions start, when nearly two-thirds of managers admit guessing rather than missing a deadline and AI reviews are overturning pay calls.
Download
Guide
Ready-made prompts for performance review language and sensitive conversations, for the 40% of HR teams already using AI to prepare or review performance assessments.
Download
Guide
When AI keeps surfacing people managers overlooked for promotion, the gap is in how leadership potential is spotted and developed, which is exactly what this guide audits.
Download
Employers that cut for AI first and analyzed later are now negotiating with the people they let go.
HR's reputation as a strategic function has outrun its operating model.
AI has changed workforce planning, but few CHROs have built it into enterprise-wide plans.