Companies are getting stuck with AI – the problem isn’t talent, it’s leadership

AI is not failing in financial services because of a lack of talent. It is failing because leadership teams have not defined how much risk they are willing to take.

That is the key finding of a new analysis from PA Consulting and Copenhagen Fintech, published today, based on interviews with senior leaders across the Nordic financial sector.

Companies are investing heavily in artificial intelligence. Expectations are high, and investment is keeping pace. Yet many organisations fail to move beyond the pilot phase, and the value never materialises.

The explanation is often framed as a lack of talent. That is a real challenge. There is strong competition for engineers, AI architects and people with business profiles who can turn technology into practice. But that is not where progress truly stalls.

The biggest barrier is that leadership has not defined which risks the organisation is willing to take, nor established clear frameworks for how decisions should be made within those boundaries.

“It is critical to define risks and risk appetite at a strategic level, so you don’t have to start from scratch with every single solution,” says Ida Bach, one of the contributors to the analysis.

When direction is missing, progress stops.

Generative AI intensifies the challenge. Where traditional AI worked with structured data, generative AI works with text, code and complex relationships. That increases both opportunity and risk. Many organisations respond by slowing down, testing cautiously or limiting themselves to simple use cases. Some stop altogether as uncertainty grows. This is not about capability. It is about lack of clarity.

When risk appetite is not defined, every new solution becomes a new negotiation. Every decision has to start from scratch. That creates friction, delays and ultimately stagnation. Effective governance does the opposite. It sets clear boundaries, enables faster decisions and gives organisations the confidence to move from pilot to production.

At the same time, many organisations still treat AI as a standalone technology project. That is a misreading. AI affects the entire organisation – from compliance and security to operations and reputation – and should be managed as a cross-cutting risk, on a par with ESG and cyber security. That requires it to be anchored at executive and board level, with clear ownership across the organisation.

The organisations that succeed with AI do five things consistently:

They define risk appetite at executive level. Risk appetite is set strategically, not project by project. Use cases are classified by risk level, with clear boundaries for what is acceptable. This means the organisation does not have to start over again every time a new solution emerges.

They establish governance that drives progress. Clear decision pathways, standardised assessments and transparency across models and data replace ad hoc decisions. This enables fast action without losing control.

They anchor accountability at the top. AI is treated as a leadership responsibility, not an isolated IT initiative. Ownership is clearly defined across business, compliance and technology, with a direct link to executive management and the board.

They use AI as an opportunity to fix the foundations. Conflicting documentation, outdated processes and poor data quality are addressed before scaling. This is rarely the most visible work, but often the most critical.

They maintain consistent pressure in the chosen direction. Setting the direction is not a one-off decision. It must be sustained through leadership, a clear focus on outcomes, organisational engagement and active culture-building. Without it, uncertainty quickly emerges and organisations begin to move in different directions. You can think of it like a river. What matters is not how fast the water flows, but whether the direction is well defined. Without clear banks, the water finds new paths. With clear guidance, it follows the intended course.

AI is not failing because organisations lack talent. It is failing because organisations lack clarity on risk, direction and accountability. When leadership takes that responsibility and maintains momentum, AI can move from pilot to production and begin to deliver real value.

Read the article in Finans in Danish.