Two experts say HR leaders will demand proof from AI vendors and rethink hiring as entry-level roles vanish
HR leaders heading into 2027 face employees who distrust polished AI output, candidates who will judge every step of the hiring process on its relevance, and a leadership bench thinned by disappearing entry-level jobs.
Those are among the 2027 workforce predictions from two specialists at The Predictive Index (PI): Anthony Belluccia, an industrial-organisational (I/O) psychologist, and Matt Poepsel, the firm's vice president of talent optimisation. Both argue that the next phase of AI at work will be shaped less by the tools themselves than by how people respond to them.
Belluccia says AI fatigue is already changing how people read each other's work.
"AI-generated content is on every platform in every format, and people are so fatigued by it that they scan for signs of AI before reading what's in front of them," he said. "In 2027, that fatigue carries into the workplace, where colleagues and managers will start trusting work with noticeable mistakes more than work that looks perfect."
He points to research on "workslop" – AI-generated work that looks finished but fails to move a task forward. In a September 2025 survey of 1,150 full-time US desk workers by BetterUp Labs and the Stanford Social Media Lab, published in Harvard Business Review, 42 per cent of respondents said they saw colleagues who sent workslop as less trustworthy. HRD has also reported new research showing AI workslop puts leadership trust at risk inside organisations.
"Next to all that polish, a typo or a rough edge becomes a fingerprint of somebody struggling through the problem," Belluccia said. "Managers will put more faith in teams that feel safe experimenting and sharing imperfect work, and those teams will do better work because problems and solutions surface early instead of hiding under AI gloss."
AI in hiring: candidates want relevance, HR wants proof
Belluccia expects candidates to accept AI in recruitment – on one condition.
"In 2027, candidates will accept AI in hiring as long as every part of the process clearly ties back to the job they're applying for," he said.
He cites the Greenhouse 2026 Candidate AI Interview Report, released in May 2026, which surveyed 2,950 active job seekers across the US, UK, Ireland, Germany and Australia. Only 19 per cent said they wanted less AI in hiring. "Candidates could forgive a hard test, but they won't forgive an irrelevant one," Belluccia said.
Relevance is also a moving target. The World Economic Forum's Future of Jobs Report 2025 found employers expect 39 per cent of workers' core skills to change by 2030.
"In 2027, I expect job descriptions to become living documents that get updated weekly as AI rewrites what each role needs," Belluccia said. "If you're still measuring candidates against last year's job description, you're hiring for a job that no longer exists."
Poepsel predicts HR will turn a harder eye on the vendors behind these tools.
"2027 is the year HR starts checking the receipts on the AI tools it uses to hire, promote and evaluate people," he said. "HR teams will ask vendors for proof that these tools work as promised and don't discriminate, and I don't expect many vendors to have a satisfying answer. That's not because they're hiding anything. They're racing to keep up with competitors, and little of their roadmap goes to safeguards."
The return on AI in recruitment is already under question, with the 2026 ELMO benchmark finding hiring takes longer despite AI adoption – 25 days to fill a role, up from 20 days in 2025.
Poepsel also describes a 2024 case in which an HR software company announced it would give AI agents employee records, then reversed course three days later after public backlash. That matches San Francisco-based platform Lattice, which dropped its "digital workers" plan in July 2024.
"People resist AI when they don't feel safe, and how they resist depends on how they're wired," Poepsel said. "Some push back openly. Others quietly stop using the tools, work around them or feed them bad data. Managers who can't see those differences will mistake silence for acceptance."
AI agents, chief AI officers and an empty leadership bench
Poepsel warns that companies giving AI agents personalities are building the wrong ones.
"Most of those personalities are being built to give users exactly what they want, which produces the silicon sycophant: an agent that avoids friction, agrees with everything and quietly degrades the work," he said. "Agents built to understand people will get used. Agents built to mimic them will get ignored."
He is equally sceptical of the chief AI officer role, which he expects board pressure to multiply in 2027.
"Naming someone to own AI feels like progress, because now there's a person accountable for it. But it doesn't solve much," Poepsel said. "Title proliferation isn't transformation."
His starkest prediction concerns succession. He cites an August 2025 Stanford Digital Economy Lab working paper by Erik Brynjolfsson, Bharat Chandar and Ruyu Chen. Using payroll data from ADP, the researchers found a 13 per cent relative decline in employment since late 2022 for 22- to 25-year-olds in the most AI-exposed occupations, after controlling for firm-level shocks.
"Entry-level jobs are where people learn a company's business, customers and history," Poepsel said. "AI has broken that first rung, and the savings look smart until experienced employees leave and the only backfill is outside hires who don't know the business."
HR leaders are raising the same alarm, with HR chiefs warning they must rebuild the entry-level pipeline as AI strips out the routine work that once built judgement.
Poepsel's answer is targeted investment rather than volume.
"The fix isn't overhiring. It's hiring entry-level people for the drives that help them learn fast, then reinvesting some of what AI saves into developing them," he said. "Otherwise, when the leadership bill comes due, no one will be there to cover it."
