AI is a shift, not a tool
Dr Marcel Lukas, Vice-Dean Executive Education and Senior Lecturer in Banking and Finance at the University of St Andrews Business School, leads the Executive Education programme Generative AI in Business and Finance: From Understanding to Implementation. Here, he reflects on why many organisations are misreading generative AI and why that will cost them.
We are at an uncomfortable point in the conversation about generative AI. In some boardrooms, it is overhyped. In others, it is still quietly underestimated. What concerns me is not either reaction on its own, but the combination of the two.
A tool at the surface, a shift underneath
Leaders are often impressed by what the tools can do today. They can draft a briefing in seconds, summarise complex material or generate client communications at speed. That creates understandable excitement. But it is surface-level excitement. It focuses on immediate productivity rather than on what happens when this capability is embedded across an organisation.
At the same time, many leaders are underestimating what generative AI will do to the way their organisations actually run. They are treating it as another tool to add to the stack, when in reality, it represents a structural shift. This technology does not just automate tasks. It learns, improves, and integrates with other systems. As it spreads through an organisation, it changes how quickly decisions are made, how risk is assessed and how customers are served. Over time, that alters cost structures and competitive positioning in ways that are difficult to unwind.
Why this is not an IT project
This is why treating AI as an IT project is such a mistake. The moment it becomes a purely technical conversation, it is pushed a the margins of the organisation and stripped of its strategic weight. But AI now touches core business questions. It influences how capital is allocated, how much risk an organisation is willing to take and how it plans to compete over the long term. Framing it as infrastructure misses the point. It is about how the business works and where it is heading.
When a chief executive says, “We’re experimenting with ChatGPT, so we’re covered,” I understand the instinct. Experimentation builds familiarity. It signals openness. But experimentation without governance, without clarity on where value sits in the business and without awareness of what competitors are doing is not a strategy. It is exploration without direction.
Leaders do not need to get lost in the technical detail, but they do need to understand what these changes mean for their organisation. When machines can generate, analyse, and interact at scale, decisions are made differently, risks emerge in new places, and value shifts across the business. If leaders distance themselves from that shift, they also distance themselves from the choices that shape it.
Too often, the conversation gets reduced to a question of speed. Are we moving too slowly or too quickly? The real risk is moving without clarity about what is actually changing. Organisations that rush ahead without that clarity make costly mistakes, invite regulatory scrutiny, and undermine trust. At the same time, some leaders let caution turn into delay, telling themselves they are being careful, while others quietly build capability. The real task is not choosing between speed and caution. It is engaging properly, so that action is informed and deliberate rather than reactive or hesitant.
The challenge for leaders is to stay close enough to the technology to understand its implications, while moving decisively enough to avoid falling behind. That requires involvement, not just briefings. It means asking hard questions about where value sits, how risk is managed and how the technology is actually being used across the business. Leaders do not need to run pilots themselves, but they do need to shape direction rather than observe it from a distance.
Where adoption breaks down
Adoption falters most often in organisations that layer generative AI on top of processes that were already broken. When workflows are fragmented, data governance is weak, and decision-making structures are unclear, the technology does not solve those problems; it magnifies them. Risk also builds quickly when enthusiastic bottom-up use is not matched with shared standards for quality and acceptable practice, because that is not transformation; it is unmanaged exposure.
Productivity gains are typically the entry point, and they are real. But stopping there is a strategic error. Generative AI affects how products are designed, how advice is delivered, how compliance is handled and how teams are organised. In financial services in particular, advisory models and risk functions are already evolving. The productivity framing feels safe because it does not force a rethink of the business model. That sense of safety will not last.
Governance is another area where many organisations are behind. In my experience, frameworks are either absent or drafted in haste, focused mainly on avoiding embarrassment. Governance should not be viewed as a constraint. It is what makes innovation sustainable and credible. Regulation will tighten. Organisations that invest early in strong governance and data strategy will be better positioned than those forced to react later.
The gap will widen
Over time, generative AI will widen the gap between organisations that learn quickly and those that do not, and that gap will have little to do with size. Smaller, well-led organisations can often move faster than large institutions weighed down by legacy systems and layers of committees. What ultimately matters is the speed of institutional learning, because those who build real capability with these tools will pull further ahead while others struggle to catch up.
Generative AI is, therefore, not a side project to be delegated and revisited later. It is already influencing how organisations operate and compete. The real question for leaders is not whether to engage, but how seriously they are prepared to take it and whether they are willing to treat it as a strategic priority rather than an optional experiment.
Learn more
These ideas underpin Generative AI in Business and Finance: From Understanding to Implementation, a three-day Executive Education programme helping leaders move from experimentation to structured, responsible implementation. Find out more about the programme and upcoming dates on the Generative AI in Business and Finance webpage.