Experts warn that deploying AI tools without upskilling creates rework, data risks, and a leadership disconnect
Deploying artificial intelligence (AI) tools without equipping employees to use them is not accelerating performance, it is amplifying errors, eroding trust, and widening a dangerous gap between what leaders believe and what workers experience on the ground.
According to Skillsoft's Workforce Readiness Report: AI Edition, conducted between March and April 2026 with 2,000 full-time employees, managers, and executives, 86 per cent of employees now use AI tools at work, yet only 24 per cent feel fully equipped with the skills to use them effectively.
At the same time, 77 per cent of leaders believe their organisations have set employees up for success – a 53-point gap between how leadership perceives readiness and the reality for employees.
Two experts who spoke with HRD say the consequences of that gap are playing out in organisations every day, and that HR leaders are the ones best placed to close it.
When AI lands without a foundation
Mark Onisk, senior managing director of talent strategy and transformation at Skillsoft, argues the problem begins the moment tools are introduced without adequate preparation.
"AI adoption without applied training often creates rework because organisations deploy tools without ensuring employees understand how to use them effectively or responsibly," Onisk told HRD.
"When AI is layered on top of disconnected or misunderstood data, it amplifies the 'noise,' producing outputs that require fixing."
Skillsoft's research found that only 16 per cent of employees receive training before new AI tools are introduced, meaning that training is routinely out of sync with deployment.
The report also found that fewer than one in ten employees feel their organisation has comprehensive AI governance in place – a structural gap that leaves individuals navigating complex tools without clear boundaries or oversight.
Sophie Bretag, HR and leadership specialist and author of The Kind Way, warns the risks extend well beyond rework.
"AI can quietly amplify disconnection when it's adopted without intention or proper training," she said. "One of the risks worth paying close attention to is how AI can often reduce the level of humanness in how we communicate. I've seen, and been on the receiving end of, cut-and-paste AI-generated emails and they can land as cold, lacking in empathy and sometimes, quite rude."
Bretag also flags privacy as a critical but often overlooked concern. "If people are inputting sensitive data without understanding the implications, it creates risks that could lead to serious breaches for their organisation."
Moving from tools to transformation
Onisk says the shift from experimentation to meaningful AI adoption requires organisations to stop layering tools onto existing processes and instead rethink how work gets done.
"Agentic AI enables this shift by automating common tasks and responding to external triggers, allowing workflows to operate autonomously and with greater intelligence built in," he said. "What's most important is defining the human oversight and governance of these workflows so that the outputs can be trusted by the organisation."
That governance, Bretag argues, must come before the tools – not after. "My advice to people leaders is to start with the why before anything else. What problems are you actually trying to solve, what do you want this tool to streamline or clarify, and where are the clear boundaries around how it will, and won't, be used?"
She recommends investing in training delivered by someone who genuinely understands the risks and complexities of the specific system being adopted, alongside policy and procedure frameworks that give staff clear expectations and a transparent pathway for when something goes wrong.
What effective AI upskilling looks like
Rather than treating AI as a separate training topic, Onisk says leading organisations are embedding learning into the flow of work. "The most effective programmes shift the focus from job roles to skills. By understanding the capabilities their workforce already has, and where gaps exist, organisations can deliver targeted learning that prepares employees to work alongside AI systems."
The skills that endure, he adds, are not technical ones. "The skills that matter most are the ones with no half-life: logic, reasoning, judgment. Those are perpetual and durable, and they're what will separate the organisations that scale AI effectively from those that don't."
For Bretag, communication and ongoing feedback are equally non-negotiable. "Feedback and communication will be the key to helping leaders and teams adopt AI in a responsible manner, together. When staff feel informed, supported, and genuinely heard in how they're navigating these tools, they're far more likely to use them responsibly and to speak up when something feels off."
The stakes are significant. IDC estimates that AI skills shortages may cost the global economy up to $5.5 trillion by 2026, driven by product delays, quality issues, missed revenue, and impaired competitiveness.
The organisations that will benefit most from AI are not those that deployed it first – but those that made sure their people were genuinely ready to use it.
