Radhika Raichura Co-Founder & CHRO, Helo.ai
By Radhika Raichura, Co-Founder & CHRO, Helo.ai
22% of workers globally feel confident their job is safe from elimination.
Yet companies keep telling me their AI rollouts are going fine.
I have started to find that suspicious rather than reassuring.
The numbers usually back them up. Licences are being used, adoption is climbing, and very few complaints reach HR. On paper, it looks smoother than any technology change I have seen in twenty-five years of building our company.
I think we are measuring the wrong thing. There is plenty of discussion about whether people are capable of using AI. I think there is another question we don’t ask often enough:
Do people feel safe telling us when it isn’t working?
People hold back from using new tools for all sorts of reasons. Sometimes they don’t want to look stupid while they are learning. That is a very human thing. But put that next to what people actually think about their jobs.
ADP surveyed more than 39,000 workers across 36 markets. Only 22% said, with confidence, that their job was safe from elimination. So, we are asking someone to tell us honestly whether an AI tool is working, while that same person may be wondering whether the tool is there to replace them.
Of course, that changes what we hear.
They use the tool when someone is watching. They say the rollout is going well. Then they quietly do the work twice. We measure compliance and call it adoption.
The dashboard looks good. Usage is up. Licences are active. Training is complete. But none of that tells us whether people trust the technology, or whether they have learned that it is safer not to complain.
I want to be careful here, because I am not describing other people’s companies from a safe distance. We have the same problem.
We ask for feedback on tools we have already committed to buying, and people can tell the difference between being consulted and being informed. Nobody has to say anything for that difference to register. It changes what comes back to you.
The instinct at this point is to run a survey, ideally anonymous.
I have stopped believing this is much of a fix. A survey is one more thing people answer carefully. And anonymity does not help much when the team is small enough that everyone knows where an answer came from.
You cannot solve a trust problem with a better way of collecting opinions.
The question I have found more useful is simpler:
Can you name someone in your organisation who openly said an AI tool was not working, and felt safe enough to say it?
More importantly, did anything good come from speaking up?
Did they gain something from being honest? Did others see that honesty was valued? Did anything actually change?
If the answer is no, then people have probably learned that staying quiet is safer.
And if nobody complains during a period when so many people are worried about their jobs, I don’t think we should automatically read that as reassurance.
It may be the opposite.
None of this is an argument against the technology. We use it. It does useful work, and I would not go back.
It is an argument about what we are entitled to conclude from a quiet rollout.
The uncomfortable part, for those of us in HR, is that this sits with us rather than with the technology team.
Whether people can afford to be honest is really a question about the culture we built long before any of this arrived.
AI did not create that condition.
It has just given us a very clear test of it.
I do not have this fully worked out.
What I am reasonably sure of is that a rollout nobody complains about, in a workforce this anxious, deserves more scrutiny than it usually gets.
(Views expressed in the article -‘The Trust Problem Behind AI Adoption‘ are solely of the author.)
About The Author
