How to make your workforce adopt AI

Even though AI has the potential to improve productivity, these gains don’t scale automatically across the economy or even within an organization. There are two underlying causes for this, according to our research.

1. Different jobs have different tasks and different tasks have a different AI exposure

AI acts on tasks, not job titles. Work that involves tasks such as analysing, writing and summarising is highly exposed to AI. Manual, physical or on-site service tasks, on the other hand, are less affected. So, some employees face meaningful opportunities to use AI immediately, while others do not.

2. Different jobs have different incentives, or ‘pushes’ to adopt AI

Incentives for people to use AI are the rewards, pressures, or the push people feel. For adoption to happen, the motivation to change must outweigh the effort.

Real organisational productivity improves only when employees learn to use AI safely, effectively, and consistently. That requires new skills, redesigned processes, governance, and trust. These are costly investments that are not captured in normal productivity measures, but they act as essential internal incentives. By providing this support structure, organisations lower the barrier to entry and make adoption attractive, effectively substituting for external pressure when it is missing.

This investment is particularly crucial because the natural ‘push’ to adopt AI varies wildly across the workforce. Employees do not all face the same labour market pressure (how strongly the competitive incentives are for workers in a given occupation to change the way they work). Some occupations operate under intense competitive pressure: high turnover, low job security, weak protections. In these roles, workers are more motivated to adopt new tools that improve productivity or safeguard their jobs. Other occupations are typically embedded in formal institutions, with strong professional norms, clear hierarchies and, in many cases, regulated entry and tenure protections. These employees feel less external pressure to change how they work.

The key insight of PwC’s research is that labour market pressure increases AI adoption not when AI exposure is very low, and not when it is extremely high, but in the broad middle where AI is relevant without yet being inevitable. When AI exposure is too low, no amount of labour market pressure can compensate for the absence of technical capability – think of cooks or dishwashers. When AI exposure is very high, by contrast, the benefits of using the technology are so evident that workers adopt it regardless of labour market pressure – that is the case of computer programmers, for example. The ‘push’ from market forces matters most in the middle of the distribution – precisely where many organisations are struggling to move from pilots to widespread, everyday use.

To illustrate the decisive role of pressure in this 'middle ground', compare nurses with chemical technicians. Both occupations have average AI exposure – meaning the technology is useful but requires effort to integrate. However, their adoption rates diverge sharply due to differences in external incentives. Nurses typically face low labour market pressure, often shielded by strong institutional norms and high demand for human care; consequently, their AI usage remains low. Chemical technicians, despite having similar technical exposure to AI, operate under significantly higher market pressure to optimize efficiency. In response to this competitive 'push,' their adoption of AI tools is notably higher.

This struggle to operationalise AI in the 'middle ground' is not just theoretical; it is reflected starkly in recent employee data. ‘Our yearly Hopes & Fears Survey shows a striking result', says PwC’s Marlene de Koning, expert workforce transformation & technology at PwC. She refers to PwC’s global study on employees’ experiences and expectations regarding their jobs.

‘More than half of employees do not integrate generative AI tooling into their daily work, even though many organisations make the tools available. But at the same time, I’m starting to see my clients move beyond viewing AI purely as a technology. Instead of treating it like just another system to implement, they’re beginning to recognize that its real impact comes when people are adopting and engaging. AI isn’t just about tools – it’s about transforming how teams work and about unlocking outcomes like greater efficiency, higher quality, innovation, and improved employee satisfaction.’

But how to make your workforce adopt AI? ‘Organisations now often take a kind of one-size-fits-all approach to AI. Everyone receives the same training', says De Koning. ‘But it would bet better for organisations to tailor their strategy to the level of AI exposure and level of labour market pressure per task. To that end, we have developed the AI Mind Map, a tool to measure and map attitudes that influence AI adoption. Basically, it measures the willingness to change. It helps organisations understand and act on the attitudes, motivations, and perceptions that influence AI adoption at the individual, team, and organizational level. It also provides a complete view of the AI mindset within your organisation, which provides data-driven input into your AI adoption strategy. And it allows you to track progress over time, thus supporting continuous insights in room for improvement.'

Once you have established the level of ‘AI-readiness’ of your workforce, you can develop a plan to promote adoption. In doing that, creating a ‘psychological safe space’ is very important, says De Koning. ‘As a leader, you must convince people that it’s OK to experiment and that they can make mistakes without being punished for it. They now must trust algorithms. That's just psychologically more difficult than trusting a human being.'

1. Make AI literacy practical and role-specific

Generic training is ineffective. AI literacy must focus on:

• Concrete tasks AI can support in the role.
• Examples from real workflows.
• Safe, low-risk experimentation.

2. Align incentives with desired behaviours

Incentives can include: 

• Recognition for effective, responsible AI use.
• AI skills reflected in performance and growth.
• Time and space to experiment with AI.

3. Build trust through governance and transparency

Employees adopt AI when they trust:

• Data handling, privacy and guardrails.
• Clear guidance on when to trust and verify AI.
• That AI supports rather than replaces their role.

4. Equip managers to lead adoption

Managers need:

• Their own AI training and confidence in using it.
• Support to redesign workflows with AI in mind.
• Simple tools and examples of what ‘good AI use’ looks like.