The AI Paradox at Work: What 8,800 Workers and Leaders Just Revealed

Hi, everyone. Alicia here from Employment Hero. The story you've been told about AI at work is that either it's an economic miracle or a workforce apocalypse. Our research suggests that it's neither and that the truth is far more human and far stranger than either headline.

We surveyed nearly eight thousand eight hundred business leaders and workers across four countries, and we found something we're calling the AI paradox. So here's the paradox in two numbers. Ninety seven percent of AI adopting businesses say that they've seen at least one measurable improvement, and seventy seven percent of AI using workers say that they've hit at least one AI cause problem. Both true at the same time in the same companies. To unpack what's actually going on here, I'm joined by Lisa Cowie, our senior insights and research lead who ran this study. Lisa, welcome.

Thank you, Alicia. I'm excited to talk about this large scale study looking at both sides of the equation. So we surveyed employers and employees, and the most interesting finding wasn't a single number. It was the gap between two stories being told about the same tech.

Okay. And let's start with the name. What is the AI paradox in one clear idea?

Well, AI in the same organization is producing two very different and opposing experiences at same time. So measurable gains for the business, but real friction among some people doing the work. There's optimism at the top and anxiety on the floor. The paradox is that both are genuinely true and most coverage just focuses on one half of that. So this is not actually about employers versus employees, but recognising what's happening on the ground.

And this isn't like some companies win, some lose. It's actually the same company experiencing both at the same time.

Yeah. And there's a difference and that's pretty well reported between small and large organisations, although that could close with this tech. But the key insight isn't really that it's about large versus small. It's more what's happening within a company.

So that's what makes it a paradox rather than just a mixed result. And I think what we can sort of sum this up with is that adoption has run ahead of maturity.

Okay. And let's talk about the loudest headline first, that AI is removing these entry level jobs, but your data pushes back on that. What did you actually find?

And yeah, that's pretty heavy in the media actually. And we did find that sixteen percent of the companies that we surveyed have cut entry level roles. And among those AI was the most cited reason. So that fear isn't imaginary, but the bigger picture is what's happening at the other end that could be indicative of what's to come. So overall, we found that forty two percent of companies grew their entry level headcount over the last two years.

But among businesses with AI at the core of their operations, that jumps to sixty two percent, more than double the thirty percent rate of non adopters. So it's actually most the most AI mature firms are the ones who are most likely to see a place for entry level staff.

Wow, that's a really striking reversal. But can we say that AI is causing that growth or just that the best companies are doing both?

Yeah, it's difficult. That, you know, data like this can't indicate causation. So we can't fully settle whether AI is driving hiring or whether the most AI mature firms are just simply the strongest performers overall. So actually are more likely to hire. But at a minimum, we see that AI advanced firms are not the ones gutting all entry level roles. So this sort of AI kills the graduate job narrative might just have a bit more nuance than being absolutist.

Interesting. And one of the findings I keep coming back to is around hiring. So where do AI skills now sit when employers are deciding who to hire?

So soft skills still actually lead. It's not that AI skills have wiped everything else out. Work ethic is at number one at fifty six percent, communication forty six percent and quick learning forty three percent. But among all businesses, AI skills ranked sixth across thirty one percent ahead of prior work experience and ahead of holding a degree.

And then among AI advanced firms, AI skills rank first, pretty intuitive.

But what's really interesting is just the sort of rate of change that's happening. So fifty eight percent of businesses have already rewritten hiring criteria to weight against AI skills and another twenty four percent plan to. So that's the vast majority of businesses that are really rethinking how they hire.

That's so interesting, especially ahead of a degree. And we're seeing this pop up in the job ads themselves, aren't we?

We are actually our own first party platform data show that mentions of AI skills in job descriptions are up over two hundred percent year on year. And we've seen this year that Claude is the fastest growing skill term of twenty twenty six.

So that shift is definitely already showing up in job ads. But actually, the workforce aren't necessarily entirely ready. So we found from the data that six in ten workers rate their own AI competence as low to average.

Interesting. And so if AI skills matter this much and we're not seeing that in terms of how they're rating them, like who's actually teaching them? What did you find on the training and how people are upskilling?

Yeah, this is surprising. You know, we found that half the workforce say that their employer does little or nothing to help them build AI skills. So businesses haven't really got ahead of that yet, you know, for the most part. But actually most employees have accepted that as quite normal. So only thirty eight percent of businesses that we surveyed run formal AI training. Although among AI core firms, it's only two in a three.

So actually people are teaching themselves and more than half of employees report that they're learning from social media. So YouTube, TikTok, Reddit, Discord.

That's a really surprising twist in who's shouldering that burden. Yeah.

The balance overall does skew to individual ownership. So forty one percent of workers see upskilling as their own job versus twenty seven percent who put it on the employer. The rest, you know, put that somewhere in the middle. But interestingly the youngest workers, so those that are aged eighteen to twenty four, and they're entering a market now that ranks AI skills above a degree, they're one point eight times more likely than over fifty five to say that AI training is their own responsibility. So they really are taking on that burden themselves.

Yeah. Wow. And we're actually deep diving into this in our next episode on how Employment Hero closed that training gap internally. So if this is your issue, watch out for that one next.

And, Lisa, I wanted to touch on the one of the ones that really stopped me in the data was that AI feels like cheating. It stopped me in my tracks. It's the same technology being described in completely opposite terms, on whether or not you run a business or you do the work. Can you take us through that?

Yeah. So while sixty percent of employers think that their staff view AI positively, we found that forty two percent of AI using workers say it feels like cheating. I mean, that's a lot of people that are not feeling wholly positive about their AI usage at work.

And then on the other side of that, sixty five percent of employers say that AI is accelerating their business, letting them work faster. But sixty three percent of employees say that AI has created more work. So checking and verifying outputs.

So that same tech in the organization is causing quite different experiences or perspectives. And that really is the contradiction and that sort of paradox in action.

But why this guilt though? I think, you know, why would using a tool your employer wants you to use feel like cheating in that sense?

Well, there's often no permission and no shared norm in businesses. So the employees that are using those tools assume that they're getting away with something.

And actually it's a pretty core psychological issue because AI is really challenging and that strongly held notion of what work actually is. So this silence that's often happening at the top creates guilt at the bottom.

And that guilt drives the use of AI tools underground, which is actually the real risk for businesses because they lose all visibility and sort of knowledge of what's going on with their employees.

I would have assumed that the people who are best at AI would feel the most secure in their roles, but your data suggests the opposite.

Yeah, a real tension happening here and sort of pretty counterintuitive. So this is what we've sort of termed the expert paradox. So the people in the workforce who are most competent and that's self reported are more likely than the average worker to say that AI is building valuable skills seventy eight percent versus fifty nine. And they're more likely to agree that they're producing higher quality work eighty seven percent versus seventy three.

So that will make sense. They see the gains, but they're also more likely to feel anxiety about their role, fifty four versus forty percent. And they're more likely to feel exposed to layoffs. So the closer people get to AI, what seems to be happening is the more clearly they see both the promise and the gains that it's given them, but also the threat of having that tool in the first place.

And the productivity benefit isn't landing evenly either.

Indeed, the dividend is skill dependent. So eighty seven percent of expert users say that AI has freed up time and that time is given to high impact work. But among beginners, that's just thirty five percent. So a really significant difference based on how skilled that worker is.

And nearly three in four workers overall say that AI has boosted productivity, but only one in four say that boost is significant.

And the time saved doesn't always become free time. Only twenty one percent say it lets them work fewer hours while just as many say it fills up with more of the same tasks.

And for twelve percent, AI saves no time at all for those people. So this isn't really all or nothing. There's this sort of messy middle ground that's happening.

Very fascinating. And one of the things I think that we picked up as well was that the paradox looks very different depending on where you work. So we survey four markets, Australia, New Zealand, the UK, and Canada, and the paradox shows up everywhere, but it looks very different in each What stood out region to region?

Yeah, you're right. The tension is universal and that came through very strongly from the data. There's not, you know, this sort of huge difference by market, but the texture on the ground is actually quite local. So first of all, with Australia, we found this to be the confident but conflicted market.

So Australian workers lead on skills perception. Sixty one percent say that AI is helping them build more valuable skills, the highest of any market. Seventy five percent say it's boosted productivity, again, the highest of all the markets that we studied. But Australia also has the highest anxiety.

Forty five percent said that AI makes them feel replaceable.

So the more they seem to value it, the more uneasy they feel.

So it really Australia's really feeling that upside and the unease more intensely than anyone else.

Yeah. High conviction, high anxiety.

But also note that in Australia, workers are the most likely to report that the time they save with AI is redirected to higher value strategic work. So the benefits are really being felt there too. So onto New Zealand, this is the calmest market. New Zealand has the lowest anxiety, just thirty four percent said that AI makes them feel replaceable versus forty five in Australia. And they also have the lowest amount of AI guilt.

And New Zealand workers are also the most likely to say that AI training is their employer's job rather than their own. So a more collective mindset. But these stats are likely driven by overall adoption, which is lower too versus all the other markets that we studied.

So a lower anxiety, but also a little bit further back adoption.

Yeah, absolutely. And New Zealand tends to measure the lowest on all the measures related to AI integration and usage. But that just means there's more runway ahead for that market. Okay, the UK, this is the leaders market. The UK is furthest ahead on the business side. Seventy percent of firms are increasing AI investment, the highest of all the markets studied, and fifty five percent strongly agree that AI is driving innovation. And then on the worker side, we found that daily AI usage has jumped from just nine percent to twenty five percent in a single year.

And UK employers are the most optimistic about how their people feel. But that optimism may be running a little bit ahead of reality because guilt is still high among employees and it rises significantly among Gen Z, where over half feel that guilt.

So the UK's leadership gap between how confident leaders feel and how workers actually feel is particularly wide.

It is. And actually, the UK's got the widest optimism at the top versus friction on the floor split. And lastly, Canada. This is what we've called the guilt and shadow AI market. So Canada has high AI guilt similar to other markets, forty three percent of workers and rising to the highest level among Gen Z, fifty six percent.

And just under half of businesses say that they know their staff are using personal AI accounts at work. So this is the clearest shadow AI signal in the data. And Canadians also the most likely to say that time saved with AI just fills up with more tasks.

Canada's really bringing that shadow AI problem to the surface with nearly half using personal accounts at work. I'm sure the security team and all these businesses is not particularly happy.

Agree. Yes, it is the sharpest illustration of where there's no permission or standard. And then when that happens, as we discussed earlier, AI use doesn't just stop. It just goes off radar.

So whichever market you're in, I guess the lesson rhymes is that the gap between leaders and workers is real. It's just measured differently.

Yeah. The same paradox exists with the same tensions bubbling away everywhere. So the fix is actually similar or the same everywhere, visibility, permission, and creating a capable and confident workforce.

So let's end on the opportunity because there's a lot of genuinely hopeful findings here. AI as the equalizer. What does that mean?

AI can be the leveler. So we found in the data that thirty nine percent of small businesses report cost savings from AI, That's almost identical to the forty one percent in large enterprises.

Productivity also scales with how well people use the tools, not just with company size. So all this indicates that it really can create a level playing field for the larger and the smaller businesses together.

And for SMEs, that's a chance to compete on terms that used to only belong to the bigger players. So the conversation really has moved from adoption to competence.

Okay. So if a business leader or owner is watching this and feels behind, what are the first three things they should do on Monday morning?

So three quick things on Monday. Close the visibility gap. So find out how your people are using AI without judgment so it stops being hidden. Number two, give explicit permission and a shared standard so that AI stops feeling like cheating. Number three, invest in capability, not just tools, because it's clear from the data that that dividend goes to the skilled and not the equipped.

This will reduce the anxiety and the fear once people are genuinely empowered and supported.

Lisa, thank you so much. A big reframe for me is that AI at Work isn't really a technology story. It's a story about the gap between how it feels to lead the change and how it feels to live it. And the businesses that close that gap are the ones that really win.

If you want a full picture, have a look at the AI paradox at work report on our website. We'll have the link in the description, every stat we talked about today, and a lot more, including the complete country to country breakdowns. And if you wanna see how Employment Hero helps teams build an AI ready culture, closing that gap, exactly what Lisa described, We'll have a link to that next episode as well. Thank you, Lisa, and thank you for everyone watching. We'll see you next time.