Do you feel threatened by AI? You would be in good company. AI is coming for at least 40% of jobs globally, the number rising to 60% in developed nations. It’s embedded into every strategic conversation and the focal point of boardroom urgency. The World Economic Forum (WEF) found that at least 6 out of every 10 workers need upskilling and reskilling before 2027 or risk being left behind in the next wave of digital transformation, yet only half have access to adequate training opportunities, and the other half aren’t much better off.
Many organisations have opted for the age-old tactic of “let’s throw money at the problem and see what happens”, triggering a wave of AI academies, prompt engineering bootcamps, and mandatory “AI 101” modules. However, this approach isn’t getting them the results they projected either. This is because the employee experience differs massively from the strategic narrative crafted at the top. The workforce feels burdened by the pressure of their current workload, compounded by the need to stay abreast of AI lest they lose their jobs.
The Complex Relationship Between AI & Training
Corporate training has long had a “stickiness” problem. Harvard Business Review famously described leadership development as “the great training robbery”: companies spent around $356 billion globally on training in 2015, yet most programmes failed to shift organisational performance because people reverted to old behaviours once they returned to their usual environment.
The same pattern is now playing out with AI:
- McKinsey shows that organisations with leading digital and AI capabilities are delivering 2 to 6 times higher total shareholder returns than laggards, but closing that gap requires sustained skill building, not one-off events.
- At the same time, many employees say they lack the skills to use generative AI safely and effectively, and expect their organisations to provide that support.
All of this indicates the intent is there, the investment is available, and the urgency is real, but the training itself isn’t taking root. If the workforce wants to learn, and organisations are willing to spend, why aren’t we getting results?
To understand the gap between organisational investment and real behavioural change, we need to look at where AI training initiatives most commonly break down.
What Isn’t Landing and Why
Chesamel has honed in on the five most prevalent sticking points around AI upskilling and reskilling, preventing organisations around the globe from converting training spend into behavioural change.
Training is Not Treated as an Ongoing Process
The science is incontrovertible. Ebbinghaus’ “forgetting curve” shows that, without reinforcement, people can forget roughly 50% of what they’ve just learned within an hour, 70% within a day and up to 90% within a week.
Most AI training is still built around one-off workshops or dense e-learning modules. Learners leave with good intentions, then immediately get pulled back into BAU. Without spaced practice, nudges or on-the-job application, knowledge decays exactly as the research predicts.
Learning Modules are Generic, Yet Work is Specific
WEF highlights that “AI and big data” skills are among the most in-demand capabilities by 2027, alongside analytical and creative thinking. But employees don’t work in abstractions; they work in Salesforce, Figma, SAP, Excel, Miro and email, to name a few.
If AI training doesn’t map clearly onto real workflows (“here’s how your finance team can use this to reconcile faster”, “here’s how a project manager can summarise stakeholder updates”), it gets mentally filed under “interesting, but not relevant”. AI training needs to include practical applications tailored to the workforce’s needs, or it’s a waste of time, manpower and resources.
Conflicting Messages – Is Your Job at Risk?
The story and syntax around AI is highly emotive and leans into extremes, hovering around massive potential disruption, colossal job losses, and total workforce replacement. When employees are asked to embrace tools that are simultaneously described as existential threats, it’s unsurprising that they respond with anxiety, not curiosity. Cognitive load goes up, and training becomes something obligatory to survive, not a capability to explore and integrate into their workload.
Reuters recently reported a Deloitte survey where 68% of executives see moderate to extreme AI skill shortages amongst their teams, while IBM data suggests around 40% of workers will need reskilling within three years. These numbers do little to quell fears around job insecurity and mass unemployment.
There Is Little Incentive to Improve
Deloitte’s 2024 Global Human Capital Trends argues that work is becoming “boundaryless”, that is, no longer defined neatly by jobs, locations or traditional HR processes, and calls for a shift from narrow productivity metrics to broader measures of human performance.
Yet many organisations still measure success by hours worked and tasks completed, not by experimentation, learning and value created with AI. If a manager quietly discourages employees from “wasting time” trying a new AI workflow on client work, the strongest signal in the system is: don’t bother.
Skill Lenses are too Narrow
Upskilling initiatives often fixate on tool-specific skills like “how to write prompts for X platform”. But McKinsey’s latest research on AI and automation suggests that over 70% of today’s skills are transferable across both automatable and non-automatable work; AI shifts where and how those skills are applied, rather than making them obsolete.
Curiosity, leadership, social influence and lifelong learning are becoming just as critical as technical AI skills, and when programmes ignore these human capabilities, they miss the very muscles that make AI adoption resilient. Addressing these barriers requires rethinking not just the content of training but the underlying systems that support how people learn, apply and sustain new behaviours.”
Four Ways to Pivot
Overcoming AI upskilling fatigue isn’t about more content. It’s about re-designing how learning, work and change fit together. This means a shift from delivering one-off training events to building a learning ecosystem that reinforces new behaviours over time.
Get Specific
Start with tailored learning applications, not generic stock content. Instead of “AI for everyone” courses, anchor training around critical journeys and use-cases: closing month-end faster, responding to RFPs, personalising marketing at scale, or triaging customer queries.
The WEF highlights how AI can be used to map “skill adjacencies”, identifying where existing capabilities can be extended into new roles. Use that insight to design pathways where people can see: “Here’s how my current strengths translate into AI-augmented work.”
Be Memorable
Design for the forgetting curve, and accept that people will forget, and build your programme around that reality. Research demonstrates that retention improves with spaced repetition, regular refreshers and on-demand access.
Practical moves include:
- Short, repeated micro-lessons instead of single long sessions
- In-tool prompts and checklists (“Try using the AI assistant for this step”)
- Follow-up challenges two days, one week and one month after training
In other words, treat AI capability as a product with a lifecycle that requires regular maintenance to perform optimally.
Redefine Metrics
Shift from “AI training hours” to AI-enabled performance. If you measure learning solely by attendance or course completions, that’s what you’ll optimise for. Deloitte points to a growing shift toward measuring human performance like creativity, problem-solving, and adaptability being supported by AI rather than replaced by it.
Define a small set of outcome-based metrics: reduction in time to complete key tasks, increase in proposal throughput, improvement in customer response quality. Then align recognition, performance conversations and leadership narratives to those outcomes, not to tool usage in isolation.
Build Community
Community creation should be an integral part of any AI training course. McKinsey highlights that winning organisations remove barriers like time and access, and create environments where people can build digital skills together, rather than consuming content alone.
Set up AI “guilds” or communities of practice where cross-functional groups share prompts, successes and failures. Empower local champions inside teams who can translate central guidance into context-specific patterns. This turns AI upskilling from a top-down initiative into a grassroots movement.
Chesamel’s Approach
AI upskilling fatigue isn’t a sign that your people are unwilling to learn. It’s a sign they’re saturated with training that’s disconnected from their reality, unsupported by systems and framed in a language of fear.
The organisations that will pull ahead are those that:
- Treat AI learning as an ongoing change journey, not a curriculum drop
- Integrate training into real work with clear value stories
- Align incentives, metrics and culture with experimentation and human-centred performance
Do that, and training stops being something employees endure and becomes a capability they own, refine and demand more of, long after the slide decks have closed.
If you’re ready to turn AI ambition into sustained capability through a human-centred, end-to-end transformation model, we should talk.
