New research finds senior HR leaders trust agentic AI far more than the rest of their teams do
Ninety-eight per cent of HR teams globally are now using artificial intelligence, but the technology remains confined largely to minor task completion rather than work that changes how the function operates, according to Culture Amp's second annual AI in HR survey.
The research found that 52 per cent of HR professionals remain unfavourable toward agentic AI (AI systems that can autonomously plan and execute multi-step tasks) specifically, exposing what the report's authors describe as a confidence gap rather than a technology gap.
The finding lands as analysts including Gartner track a similar pattern in the local market, with HR leaders increasingly caught between rapid AI adoption and hesitation about handing over consequential decisions.
A gap between access and trust
Culture Amp's data shows the barrier to advanced AI use in HR isn't a lack of tools or training. HR executives are more likely than any other group in the function to use AI for genuinely agentic tasks, while more junior practitioners lag well behind – a split the report's authors say likely mirrors a broader pattern across organisations, where those closest to setting strategy have built more confidence in advanced AI capability than the rest of the workforce.
That gap matters because analysts including Gartner and the World Economic Forum increasingly expect HR to lead AI transformation inside organisations, tasked with redesigning how work gets done, building new capability and guiding the rest of the business through the shift. If the function expected to lead that transformation has a meaningful internal divide between its most senior and most junior practitioners, it signals how difficult closing the gap will be more broadly.
Justin Angsuwat, chief people and customer engagement officer at Culture Amp, said the findings point to something bigger than HR alone.
"It's our job as leaders to model agentic usage and show them what it can look like. We have to take them along for the ride," he said. "For organisations to move from tinkering to transformation, they'll need to actively build psychological safety and governance frameworks around autonomous AI use, not just deploy more tools meant to assist with one-off task completion."
Trust depends on transparency
The survey also found that 89 per cent of HR professionals are more likely to act on AI outputs that cite their sources – a finding Culture Amp says reinforces that closing the confidence gap isn't only about exposure to the technology, but about designing systems people can see the workings of and building the right governance framework around them.
Angsuwat said Culture Amp has approached its own AI rollout in stages rather than setting productivity targets from the outset.
"We set aside six weeks to drive confidence in AI, and we did that. So, 84 per cent of our organisation became confident using AI, and in fact only 4 per cent had negative sentiment around that," he said.
Rather than pushing for immediate output gains, Angsuwat said the company deliberately avoided "kind of systemic work shift as a result of that" in the early phase, instead giving staff room to experiment before expecting behavioural change.
He linked that caution to a wider concern about how AI reshapes trust at work, noting that senior staff often built their careers on skills that AI is now starting to automate. "People who are successful today were successful in the old way of doing things," he said, "and we need people to let go of that part of their identity to shift to a new way of doing things."
Governance first, enthusiasm secondA hard line on employment decisions
Angsuwat was clear that governance, not enthusiasm, has to carry the weight of managing that risk. Culture Amp runs a cross-functional AI governance committee spanning legal, risk, security, finance and engineering, but he said the goal isn't to default to the most cautious option on every issue.
"There's some things that we're more willing to experiment on and other things where we hold a much higher bar," he said. "For example, AI making employment decisions is a hard line that I won’t cross."
"There's some things that we're more willing to experiment on and other things where we have bright lines," he said. "Making employment decisions is a hard line that I won't cross."
For HR leaders navigating the same tension between speed and caution, Angsuwat's message is that the surge in agentic AI adoption doesn't have to come at the cost of workforce trust – provided governance and transparency are built in from the start, rather than bolted on after the fact.
