New research argues employers are handing AI tools to a job that needs a rebuild
Two closely watched workforce research teams are converging on the same conclusion – the "delayering" wave stripping middle managers out of organisations in the name of artificial intelligence (AI) has not actually redesigned the manager's job.
Dr Autumn Krauss, an organisational psychologist and Head of SAP SuccessFactors' Future of Work Research Lab, said the prevailing playbook – fewer managers, wider spans of control, an AI licence and a pat on the back – was falling short.
"I don't believe that the thesis is we're going to have less managers, increase span of control, give them all Copilot, that'll do their admin operational work and leave people managers to be more people-y," she said. "It's shortsighted. And it's not accurate with what AI can do."
Neal Woolrich, a director analyst in the Gartner HR practice, told HRD that Australian employers of every size are still working out how far AI adoption should go. "There are some employers who are well advanced in the journey, mostly in tech industries, but elsewhere it's still a struggle," Woolrich said.
Why "less managers, more AI" isn't working
Krauss's team surveyed people managers and found only one in 10 said they would be disappointed if they no longer had direct reports. "This isn't something that people managers are like clawing onto," she said, pointing to a "manager engagement premium" in Gallup's State of the Global Workforce research – managers historically more engaged than individual contributors – that "has basically vanished."
Krauss's research also found 93 per cent of managers said their job had changed in the last year because of AI, but mostly through their own initiative rather than formal redesign by their employer. "If you didn't know what your managers were doing before, you definitely don't know now," she said, describing the current state as "in the wild" rather than intentionally designed.
Where AI should – and shouldn't – take over
Krauss's team asked HR leaders to sort 21 typical manager responsibilities into what AI can do versus what it should do. "Mostly everyone agreed AI can do it all if you let it," she said. "But there's a big hesitation that should it do it." Pay, promotions and performance reviews drew the most resistance.
Yet when Krauss's team tested reactions to AI-driven decisions in scenario experiments, employees often preferred the AI-based outcome over an average human manager, particularly around promotion feedback. "It wasn't the best possible manager you've ever had," she said. "What about the average manager? The one that, when you don't get promoted, doesn't really have a great, reliable, comprehensive answer of why."
Woolrich cautioned against overcorrecting employee anxiety with vague reassurance. Organisations that communicate well tend to promise "valuable opportunities" to build AI skills rather than a specific future job title, and framing AI as a tool rather than a "teammate" reduces job-security fears, he said.
Gartner's own analysis points to AI eventually becoming a net job creator – "by about 2029, 2030" – but with significant labour-market disruption in the interim, a trajectory playing out unevenly according to recent Australian HR Institute survey data on redundancy expectations and hiring intentions.
What comes next for the manager role
Rather than fewer managers doing the same job with AI support, Krauss argued organisations should consider "untethering" managers from fixed direct reports, deploying experienced people into project-based teams as needed rather than maintaining permanent hierarchies. She also predicted managers will not necessarily be paid more than individual contributors in future structures, with reward tied to impact rather than seniority.
Woolrich expects the opposite pressure in the near term: managers overseeing larger, more generalist teams and increasingly managing "both people and technology" – a shift that will make the job harder, not easier, even with AI support.
For Australian HR teams, the message from both researchers is that cutting layers and handing out AI licences is a placeholder, not a strategy. A recent whitepaper on how Australian HR leaders are deploying AI across recruitment and onboarding suggests the organisations furthest along treat AI as a role-design question, not just a tooling decision – a theme echoed in recent analysis of why so few Australian employees have embedded AI into daily work.
