Five uncomfortable truths of AI adoption

It seems AI is high on every business agenda. Yet for many leaders, the volume of discussion can feel overwhelming and, at times, over-hyped. 

That reaction is understandable. Across industries and sectors, we see organisations investing in AI but struggling to realise the value they expected. The reasons are often not technical. They are leadership and transformation challenges. 

Buying licences is relatively straightforward. Funding technology initiatives is familiar. The harder work is helping people build new skills, redesigning workflows and embedding AI into the way work gets done. That’s where value is created and where many organisations are finding the greatest challenge. 

Here are five uncomfortable truths we regularly see in executive and boardroom conversations.  

1. Ambition can become an obstacle when organisations try to “boil the ocean” 

Many leaders are trying to solve for the future of their industry before addressing how they can improve performance next quarter or next year. There is nothing wrong with ambition. Long-term vision matters. But when all attention is directed towards a 2030 moonshot, organisations can overlook what matters most today: repeatable, measurable progress that builds confidence, capability, and momentum. 

The organisations making the greatest progress are not necessarily those with the most ambitious AI strategy presentations. They’re the ones delivering a steady flow of practical wins—small and medium-sized improvements that compound over time. 

Meaningful transformation is rarely delivered through a single centralised programme. History shows that major shifts—from industrialisation to personal computing—take hold when innovation is applied close to real business and customer problems. 

Governance remains essential. So do strong technology, data, and risk foundations. But it is unrealistic to expect technology and data teams alone to deliver AI transformation. 

AI is an enterprise-wide opportunity. Like revenue growth, talent attraction, or customer experience, it requires active ownership from the CEO, executive team, and leaders across the business.  

2. Upskilling and human change are too often underestimated 

AI does not create value simply because the technology is sophisticated. Value comes when people know how to apply it effectively, within the organisation’s context, risk appetite, workflows, and industry. 

Building AI capability is like building fitness in a gym. One or two sessions may create awareness, but sustained practice is what drives lasting change.

Scott McLiver
Partner, PwC New Zealand

Too often, AI upskilling is either absent or treated as a one-off event: a training session, a few prompt-writing demonstrations or a lunch-and-learn. That is not enough. Building AI capability is more like building fitness in a gym. One or two sessions may create awareness, but sustained practice is what drives lasting change. 

Today’s AI tools can give individuals access to powerful capabilities at a relatively low cost. But access alone does not equal adoption, and adoption alone does not guarantee value. The opportunity is not simply to teach people how to use a tool. It is to help them use AI to improve their work and your business: to make better decisions, serve customers more effectively, reduce administrative burden, and redesign critical processes. 

That requires role-based learning, practical use cases, clear guardrails, and leadership support—not just generic prompt training.  

3. A focus on measuring return on investment (ROI) can sometimes get in the way of achieving it 

Measuring value matters. But in some organisations, the pursuit of perfect ROI measurement becomes a barrier to action.  

AI is often held to a standard that other transformational technologies did not face. Organisations can become so focused on proving every benefit in advance that they miss the opportunity to learn, improve, and create value in practice. 

A more pragmatic approach is to recognise value when it is visible: 

• If AI identifies fraud that was previously missed, that creates value. 

• If it reduces customer response times, that creates value. 

• If it reduces a five-hour task to one hour, that creates value. 

The goal should not be to abandon measurement. It should be to match the level of measurement to the maturity and scale of the use case and improve over time.  

Waiting for perfect certainty can unintentionally delay adoption. 

4. The pursuit of perfect data can slow practical progress 

Data is fundamental to long-term AI advantage. High-quality, well-governed data will remain a critical differentiator for organisations seeking to scale AI effectively. However, some organisations are allowing the pursuit of a perfect enterprise data environment to delay clear, near-term opportunities. 

In practice, employees do not need to read every document in SharePoint or every historic customer record before they can do their jobs. They draw on relevant information, judgment, and experience to complete a specific task. AI can often do the same—when used with appropriate controls, access permissions, and human oversight. 

The long-term vision of a connected, trusted enterprise knowledge environment is compelling. But organisations do not need to wait for every data source to be harmonised before using AI to support employees today. 

Many valuable use cases can begin with the data and expertise already available to the business, provided they are designed responsibly and governed appropriately. 

5. The “boring” work is often where the value is 

Everyone wants the headline-grabbing transformation programme. But some of the most reliable AI value today comes from improving everyday work. 

That might include: 

• Reducing a five-hour task to one hour 

• Cutting a 90-minute process to 20 minutes 

• Clearing backlogs faster 

• Improving the quality and consistency of first drafts 

• Reducing rework, manual administration, and avoidable hand-offs 

• Helping employees find and use information more effectively 

These use cases can have a significant impact on productivity, employee experience, customer outcomes, and cost.  

AI success is less about a single, enterprise-wide leap and more about a portfolio of well-executed improvements that build on one another. That’s how confidence grows and how capability spreads. That’s how larger investments become evidence-led rather than speculative, and how AI can move beyond incremental improvement to support higher ambitions of business reinvention and growth. 

Where the biggest opportunities are today 

The most immediate and repeatable opportunities for AI value often sit at the intersection of mature AI capabilities and high-volume business processes. 

In most cases, there is little value in building from scratch what you can buy, configure, and deploy faster—often at a fraction of the cost.

Scott McLiver
Partner, PwC New Zealand

Three areas consistently stand out: 

1. Knowledge worker productivity 

2. Customer service and contact centre operations 

3. Software development and maintenance 

These areas share several characteristics: 

• High volumes of work 

• Language-rich tasks 

• Significant pressure on cycle times 

• Repetitive processes that can be improved at scale 

• Clear opportunities to support, rather than replace, human judgment 

Enterprise-grade tools are available today to support many of these use cases, with privacy, security and governance capabilities built in. The opportunity is to move with purpose: adopt proven solutions, tailor them to your business and focus investment on the outcomes that matter most. 

In most cases, there is little value in building from scratch what you can buy, configure, and deploy faster—often at a fraction of the cost.