HR doesn’t have an AI adoption problem. It has an AI trust problem

The organisations and the HR functions that make that shift first won’t just see better AI outcomes. They’ll be the ones still setting the agenda when everyone else catches up, writes Justin Angsuwat.

Ask most HR leaders whether their function has embraced AI, and the honest answer is yes, overwhelmingly. New data from Culture Amp’s second annual AI in HR communities survey of HR professionals shows adoption is no longer the story. Almost every HR professional is using AI in some form. The real story lies in how they are using it – and how they can continue to evolve and develop that skill set.

Right now, it’s mostly drafting, brainstorming, and summarising. Our data shows HR professionals are overwhelmingly putting AI to work on content creation (86 per cent), brainstorming (83 per cent) and synthesising information (81 per cent). These are the same top use cases we saw last year. Useful, certainly. But this is AI being used the way you’d use a very capable intern: as an assistant to be directed, not a partner to be trusted with real work.

Meanwhile, the technology with the greatest potential to change how HR actually operates is sitting largely untouched. Only 24 per cent of HR professionals are comfortable letting AI act autonomously on their behalf. Just 39 per cent have moved AI into HR operations automation, and only 34 per cent are using it for agentic workflow support. HR is standing at the edge of the shift that matters most, and mostly choosing not to step in.

The professionals who have crossed into agentic territory, letting AI act within boundaries they set rather than simply responding to what they ask, are seeing process transformation at rates two to three times higher than everyone else.

That comfort gap has consequences. Those who have made the leap, what our data calls “the transformed”, are nearly three times as likely to say AI has made them better leaders than those who haven’t (70 per cent versus 25 per cent). A majority are already planning to hand transactional HR work to AI altogether, compared with barely a quarter of the sceptics still on the sidelines. The upside for those willing to move is not marginal. It’s significant.

So, what’s holding the rest of the function back? Not capability, and not access. It’s trust. Our survey found 89 per cent of HR professionals say they’re more likely to act on AI-generated insights when the sources behind them are cited. Explainability isn’t a nice-to-have feature bolted onto AI tools. For HR, it’s the entire foundation on which willingness to cede any autonomy is built. You cannot ask a profession built on judgement and duty of care to hand over decisions to a system it can’t interrogate.

There is also a widening readiness gap worth naming. HR C-suite leaders are considerably more comfortable with agentic AI than the rest of their function: 37 per cent of HR C-suite leaders are comfortable with it against just 16 per cent of individual contributors, and 54 per cent of C-suite leaders are already using AI for agentic workflow support, more than double the rate of anyone else. Executives are modelling the behaviour. The function underneath them hasn’t yet followed. Left unaddressed, that gap doesn’t just stall HR’s own transformation. It puts HR at risk of losing the very strategy ownership it has been fighting to claim. The share of HR professionals who believe HR should own an AI strategy at work has grown 10 percentage points year over year, from 37 per cent to 47 per cent. That’s a real claim to influence. But a profession can’t hold strategic ownership of AI while remaining the most hesitant function to actually use it.

Where to from here?

This is the moment for HR leaders to move deliberately from task-level assistance to bounded agentic deployment, not as a leap of faith, but as a managed experiment with clear boundaries and real governance behind it. Leaders will increasingly need to manage agents and agentic workflows the way they manage human teams today, operating on the principle that agents will do real work under real authority. That means building the psychological safety and governance frameworks that make autonomy safe to grant, not simply issuing more tools for one-off tasks and calling it transformation.

I want to be transparent about where we are with this at Culture Amp. I own this transformation as the chief people officer, and I don’t take that lightly. It’s a rare thing to be leading. There are two jobs inside it: deploying the technology and changing how work gets done because of it.

HR leaders can, and I argue should, own both. I think we’re better placed than most to do so. Because there is no playbook. The work to be done is to set the principles, define the organisation’s risk tolerance, build the guardrails, and then enable people to get the most out of it.

We started with the work, not the tools. I laid out six principles for how we believe we should structure work in an AI-enabled organisation, for example, that context should be shared and machine-readable, which, it turns out, is actually good for both humans and machines. From there, we defined our risk tolerance, which is the guardrails for what we can accept and the ones we can’t. That clarity lets us scale very quickly rather than relitigate every decision.

And then, we deliberately let a thousand flowers bloom. The near-term productivity gain is pretty limited, and we’ve been honest about it. The ROI is really familiarity, people connecting their tools to the models and experimenting until the technology feels natural. Then we can build from the centre much more efficiently.

The organisations and the HR functions that make that shift first won’t just see better AI outcomes. They’ll be the ones still setting the agenda when everyone else catches up. The window to do that deliberately, rather than reactively, is narrower than most HR leaders think.

Justin Angsuwat is the chief people and customer experience officer at Culture Amp.

Want to see more stories from trusted news sources?
Make HR Leader a preferred news source on Google.