I write about using AI to do better finance work. From the CFO seat.

I heard from a colleague an analogy about the current state of AI that resonates quite nicely with what I think is happening at the moment.

Milton Friedman was once touring a construction site overseas. Workers were moving earth with shovels instead of machinery. When he asked why, the host said it was to preserve jobs. "Then why not give them spoons?" Friedman replied.

That story is 50+ years old and the logic hasn't changed. Every time a better tool shows up, the first instinct is to protect the old workflow instead of rethinking the work itself.

Consider what finance looked like before Excel. Well, obviously (and fortunately) I would not know first hand, but from what I've read and heard: spreadsheets were literal sheets of paper. Consolidation meant a team of people with calculators cross-checking numbers for days. Forecasting was a quarterly event because it physically took that long.

Nobody using Excel today can even imagine going back. But when it first appeared, the conversation was the same: "This will eliminate jobs." It didn't. It eliminated tasks. And the people who learned to use it properly became 10x more valuable than those who resisted.

AI in finance is the same story playing out again. After discussing this with a forward-thinking bank CEO, something clicked for me. The mistake most teams make is treating AI as a cost-cutting exercise. "We'll automate a few roles, save some headcount." That thinking puts a ceiling on what you can actually achieve.

The question worth asking instead: if you had to design this process from scratch today, with full knowledge of what AI is capable of, would it look anything like what you currently have? In almost every case the answer is no. Entire steps become redundant. Others consolidate. Work that used to be impossible becomes straightforward.

Getting there is not easy. Redesigning workflows from first principles requires significantly more effort than simply plugging AI into an existing seat. But that is where the compounding returns actually come from.

From what I've seen so far, AI increases quality and output per person so much that it doesn't necessarily require replacing anyone. The process redesign will come naturally once teams actually start using it and realize what's possible. And as always, the best is yet to come.

Therefore, I am not worried about AI replacing finance professionals. I am worried about finance professionals who refuse to rethink their work with AI. Those are two very different problems, and only one of them is actually solvable.