News

Pankaj Prasoon is senior director of CFO systems at LinkedIn, leading finance technology and building AI fluency across finance teams.

getty​The hardest part of AI transformation isn’t getting people to trust AI. It’s getting great people to let go of the things that made them great.

Between the world wars, every major navy could see that aviation would change war at sea. The technology was visible, and the evidence was mounting. But organizations built around the battleship did not abandon decades of doctrine, expertise and status simply because something better had arrived. The obstacle was never comprehension. It was that accepting the new model meant questioning the value of the old one.

Finance is running its own version of that transition.

We usually describe AI readiness through three lenses: mindset, skill set and tool set. I would add a fourth, dataset. All four are hard, but for entirely different reasons. Tool set is an investment problem, solved with platforms, agents and integration work. Skill set is a capability problem, solved with time to learn and experiment. Dataset is a structural problem, because today’s data quality reflects years of process decisions, system implementations, acquisitions and work-arounds, and there is no shortcut through it.

Mindset is an identity problem. You cannot solve an identity problem with another training program.

When Expertise Becomes Personal​In transformation after transformation, the people most hesitant to automate a process are typically not the weakest performers. They tend to be the strongest.

For example, take a controller who built a career on producing a clean close. She knows where problems hide, which reconciliation deserves attention and who to call when something doesn’t tie. We are now telling her that production should be automated and her value sits in governance, judgment and exception management. Or take the analyst known across the business for a model nobody else can rebuild. That complexity was evidence of expertise. Now we say the model should be reproducible and the knowledge accessible.

These people have not failed to understand AI. They understand the implications better than most. We are asking them to write down an asset they spent 20 years building, and that is a very different conversation from “AI will make you more productive.”

Why Finance Doesn’t Understand Sunk Costs In A Career​There is an irony here for a profession built on economic rationality. We will impair an asset without sentiment the moment its economics change. When the asset is our own expertise, the calculation gets much harder.

So the strongest performers tend to resist hardest, and their response is not irrational. The market value of certain skills genuinely is being repriced. The mistake leaders make is pretending otherwise. We talk about augmentation, run demos and announce training, while the question people are actually asking is far more personal: If AI can do more of what made me valuable, what makes me valuable next? Until leaders answer that, adoption will lag behind the technology.

Making Repricing Visible​Here’s how to make the repricing more visible.

Say clearly what is changing. Ambiguity suppresses adoption more effectively than skepticism. Be explicit about which capabilities are becoming commodities, which are becoming more valuable and where freed capacity goes. Nobody automates their way toward a future their leadership won’t describe.

Change what you celebrate before you obsess over what you measure. Every organization has a currency of status. In finance we celebrate whoever saved the close or rebuilt the model overnight. Those contributions are real, but we also have to celebrate the person who made sure the heroics were never needed again. What gets praised in a Tuesday staff meeting changes behavior faster than anything written into a transformation scorecard.

Become a beginner again. The strongest signal a senior leader can send is not another adoption mandate. It is using the technology, getting it wrong and showing the team what failed. When someone with 20 years of expertise is willing to look inexperienced for 20 minutes, everyone else gets permission to learn.

Give people somewhere to reinvest. It is not enough to say yesterday’s advantage is worth less. Name tomorrow’s: framing the right question, challenging an AI-generated conclusion, spotting the anomaly, knowing which exception deserves a human. AI reduces the premium on producing the answer and raises it on knowing whether the answer makes sense.

The Lever With No Budget Line​Technology can be funded. Training can be scheduled. Data can be governed. Projects can be staffed. Mindset fits none of those mechanisms. It lives in what leaders model, what organizations reward and what people believe will make them valuable three years from now. That is why it moves slowly and why it cannot be handed to the AI team, the transformation office, HR or a consulting partner.

History’s lesson is rarely that leaders failed to see the technology coming. It is that they underestimated how hard it would be to unwind the incentives and identities built around the old model. The navies that embraced carrier aviation did not do it by convincing admirals that airplanes were interesting. They changed what the institution valued. Finance leaders will have to do the same.

The information provided here is not investment, tax or financial advice. You should consult with a licensed professional for advice concerning your specific situation.

Forbes Finance Council is an invitation-only organization for executives in successful accounting, financial planning and wealth management firms. Do I qualify?