Organisational psychologist Amantha Imber says giving staff AI licences without proper training is fuelling a productivity crisis
HR leaders are handing their workforces artificial intelligence (AI) licences without the skills to use them well, and the result is a flood of low-quality, AI-generated work that is quietly eating into productivity rather than boosting it.
That is the central warning from Dr Amantha Imber, organisational psychologist and founder of behavioural science consultancy Inventium, who addressed HR professionals at HRD's National HR Summit on how to extract genuine value from AI at work.
"People either don't disclose when they're using AI or they use it poorly and produce slop," Imber said. "This AI slop avalanche – people have the AI licences but they don't have the training – is leading to a proliferation of AI slop, which is where you get a reduction in productivity."
The phenomenon, she explained, plays out in two ways: inboxes overwhelmed by lengthy, poorly crafted AI-written emails, and managers burdened with reviewing AI-generated reports and slide decks that carry little original critical thinking.
"If you're getting all these AI-generated presentations where not a lot of critical human thinking is going into it, but you as a manager are expected to review that – that is a huge increase in workload," she said. "The promised productivity gains of AI are just not there."
The context problem most HR leaders are missing
At the heart of poor AI output, Imber argued, is a failure to provide adequate context – not just better prompts. She recommended HR professionals create a personal briefing document, built by having AI interview them about their role and responsibilities, so the tool already understands their context before any task begins.
"Context is everything," she said. "For any significant project, it should have its own briefing document – what are the project goals, what does good look like, who's involved, any factors important for your AI assistant to know to do a great job."
She also flagged a cultural issue around transparency. Leaders and team members should be disclosing how and when AI has been used, she said, particularly when AI has produced a draft with minimal human input. "If I've put a prompt into AI and spent five minutes producing a 10-page report, that is an insult to ask someone to spend an hour reviewing it."
Why IT-led training is falling short
On the question of workforce upskilling, Imber was direct: the majority of AI training currently delivered in Australian organisations is not fit for purpose. Too much of it, she said, comes from IT backgrounds rather than learning design expertise, and focuses on which buttons to press rather than how to embed AI into real working lives.
"Good AI training is based on real use cases that people can relate to. It's really interactive – people get the chance to play and imagine how this can fit into my work life," she said. "HR leaders need to think about how they're training people and whether they're putting people through training that is designed with the learner in mind – training that's designed to change behaviour, as opposed to an IT person."
This is a distinction that matters as Gartner has identified AI strategy as the top priority for chief human resources officers (CHROs) in 2026, with AI productivity gains at the core of the HR operating model shift. Yet Imber's message is that technology deployment without behavioural science will not deliver on that promise.
Additional studies show that that 71% of AI users now rely on the technology to assist with their jobs, but half of users report adopting it independently, compared with just 25% encouraged by leadership – a sign that formal, high-quality training programs are not keeping pace with grassroots adoption.
The CPO–CIO partnership that actually works
Imber was equally clear on where accountability for AI transformation should sit within organisations: not with IT alone, and not with HR alone.
"AI is not a technology problem – it's a people and culture problem," she said. "The clients we see have the best success where you've got your chief people officer and your chief information officer working hand in hand together. This is not just a job for IT and it's not just a job for HR."
For those thinking about where to begin, Imber said the starting point is workflow analysis – something most employees are not trained to do. "Most people don't think like a workflow architect or a process engineer. They don't even know where to start when it comes to identifying repetitive tasks that we do every day as a team or as a function."
Once those tasks are mapped, the next step is building agentic solutions that produce consistent, reliable output – work that requires understanding how AI works, how to build and test an agent, and how to iterate. "It kind of sounds easy, but there's quite a lot of training and knowledge that needs to come with that to get those big efficiency gains we were promised with AI."
As HRD's analysis of AI's shift from automation to augmentation shows, the technology can no longer be treated as a technical initiative owned by IT, it has become a core lever for how work is designed and how skills are developed.
Today's AI is the worst AI you will ever use
Imber closed with a reframe she uses with every audience at Inventium. Whatever frustration or uncertainty HR leaders are feeling about AI right now, she said, they should hold one thought in mind.
"The AI that you're using today is the worst AI you will use in your life. Five years from now it will seem as antiquated as Clippy does now."
For HR leaders navigating the shift away from knowledge transfer toward genuine capability building in the AI era, Imber's session offered a clear message: the licence is not the strategy, and training designed around human behaviour is the difference between productivity gains and an inbox full of slop.
