How to future-proof skills as AI reshapes work

Experimental AI use, what the data shows

The survey highlights a clear pattern, around 45% of organisations report that their use of AI is still ad hoc or experimental, while only a small minority have fully integrated it into core operations. 

However, this experimentation is often taking place without the organisational foundations typically associated with effective technology adoption. 

Many organisations report weaker capability across: 

• governance and oversight of AI use
• supporting infrastructure and systems
• innovation culture and approaches to experimentation, and
• workforce transformation planning. 

This is reflected in the survey data: around 42% of organisations experimenting with AI fall into the lowest organisational readiness group, meaning they lack strength across these core capability areas. 

Source: The survey was conducted on behalf of the CIPD by YouGov, between 14 January and 3 February 2026. The total sample size was 1,342 people professionals and business leaders.  

Taken together, this suggests that experimentation is frequently happening ahead of the organisational conditions needed to support it. Without these foundations in place, there is a greater risk that AI use remains fragmented, poorly governed or fails to deliver meaningful improvements. 

Where AI is already changing work and how organisations are responding  

Alongside these broader capability gaps, the research highlights a more specific challenge: how organisations manage the impact of AI on work and skills. 

Evidence from the research shows that AI is primarily reshaping work within roles, rather than leading to immediate large-scale job loss. Tasks are changing, skill requirements are evolving and workflows are being reconfigured. 

However, organisations vary significantly in how they are responding to these changes. Those further along in their adoption journey are significantly more likely to report activity in key areas of workforce transformation, particularly: 

• job redesign
• reskilling strategies, and  
• structured approaches to skills intelligence. 

Among organisations where AI is strategically embedded, over 80% report both job redesign and reskilling activity. By contrast, organisations still experimenting with AI are far less likely to have these in place. 

This points to a clear divide. Organisations that move beyond experimentation are not just using AI more – they are more likely to be adapting work and skills alongside it. 

The challenge is not simply adopting AI but building the capability to manage its impact on work, jobs and skills. 

The risks of skipping a staged upskilling approach  

If organisations do not take a more structured approach to AI skills planning, several risks are already emerging. 

Skills erosion and weaker learning pathways

As AI takes on more routine tasks, there is a risk that employees —particularly those early in their careers — have fewer opportunities to build foundational skills through practice. Over time, this can weaken capability rather than strengthen it. 

Work intensification rather than better work

While AI can reduce time spent on certain tasks, this does not automatically lead to improved job quality. In many cases, expectations around output increase, meaning employees are required to do more rather than differently. 

Disrupted career pathways

Changes to task composition can make roles, and progression routes, less clear. Without deliberate planning, these risks create gaps in development pathways and limit progression opportunities. 

A continued ‘efficiency trap’

Organisations may focus on short-term time savings rather than longer-term improvements in performance or job quality. This can limit the potential benefits of AI adoption. 

Without a clear approach to skills planning, there is a risk that AI is implemented in ways that weaken capability, intensify work and disrupt progression rather than improve performance. 

What a staged AI upskilling model looks like 

Skills planning should align with how AI is being used

A common mistake is to treat AI skills planning as a one-off exercise, often focused on large-scale training or reskilling programmes. However, effective AI skills planning needs to align with where an organisation is in its adoption journey. 

As CIPD’s practical guidance highlights, AI adoption typically follows a staged pattern — from early experimentation, through operational pilots, to more strategic integration.  

Each stage brings different workforce challenges and requires a different response. 

Early stage: build confidence and safe experimentation

At the early stage, AI use is typically informal and exploratory. Employees may already be using tools in their day-to-day work, often without clear oversight. 

The priority at this stage is to move from uncertainty to informed experimentation. This means focusing on: 

• creating safe conditions for use, including clear guardrails
• building baseline AI literacy and awareness across the workforce, and  
• identifying where AI is already being used and by whom. 

This stage is less about advanced capability and more about developing a shared understanding of what AI can and cannot do and establishing the foundations for more structured use. 

Intermediate stage: focus on applied use and quality

As organisations move into more structured use, the focus shifts. AI is now being used in pilots or specific functions, and the challenge becomes ensuring it is used effectively and delivers value. This requires developing: 

• task-specific skills (for example, prompting and verification)
• the ability to identify and correct errors
• stronger approaches to quality control and risk management, and  
• clearer links between skills development and business outcomes. 

At this stage, skills planning becomes more structured but remains closely tied to real use cases and operational learning. 

Advanced stage: align skills with workforce transformation

At more advanced stages, AI is embedded in how work is organised and delivered. The focus shifts from individual use cases to broader organisational change. 

This is where skills planning connects directly to: 

• job and role redesign
• workforce planning and transition pathways, and  
• organisational structure and career development.  

At this stage, organisations need to take a more strategic and system-level view of how work is evolving. This includes making more fundamental decisions about how roles change, how teams are organised, and how people are supported to transition into new forms of work. 

Skills planning becomes an integral part of wider organisational transformation, rather than a standalone activity. 

How HR can build AI upskilling in stages 

The key message for HR is not to do everything at once. 

The evidence suggests that AI adoption is often moving faster than the systems needed to support it. Skills planning needs to catch up, but in a structured and proportionate way. That means: 

• starting with where the organisation is now
• building capability through real-world application
• scaling skills planning alongside adoption. 

In other words, the question is not simply: 

'What skills do we need for AI?' it is 'What skills do we need next, given how AI is actually being used in our organisation?' 

Conclusion 

AI is reshaping work in ways that are often gradual but cumulative, changing tasks, roles and skill requirements over time. 

Organisations that respond effectively will be those that align their skills planning with how AI is being adopted in practice, rather than relying on one-off or overly broad interventions. 

For HR, this means taking a more dynamic and staged approach — one that builds capability progressively and supports organisations to move from experimentation to effective workforce transformation.