Stop Going to AI Conferences. Start Riding the Bike.
I haven't been to a single AI conference.
Not one. While colleagues have been booking seats at summits, paying registration fees for full-day events, and collecting lanyards with printed agendas, I've been sitting at my desk building things — badly at first, then less badly, then pretty well. A portfolio tracker. Automation tools. Finance workflows that I've described elsewhere on this blog. All of it learned through what I'd call the least elegant method imaginable: watching a few YouTube videos (including one from a neuroscientist at KAIST that I'll come back to) and then just starting.
I want to be direct here, because I think there's a genuinely unhelpful pattern in how finance professionals are approaching AI right now. The pattern is: attend conference → feel informed → return to desk → change nothing. Repeat next quarter. The conferences are good. The speakers are often excellent. But if you leave without fundamentally changing what you do at your keyboard on Monday morning, the conference did nothing.
This is my view. Other people's experience will differ, and I respect that. But here's the model of AI learning that has actually worked for me — and why I think finance professionals are uniquely positioned to do it well.
The Neuroscientist Who Explained It Better Than Anyone
Professor Kim Dae-sik is a neuroscientist and professor at KAIST — one of Korea's leading technical universities. I came across his perspective on AI learning while watching content about how people actually acquire new capabilities, and his framing stuck with me immediately.
His point: learning AI is like learning to ride a bike. You cannot learn to ride a bike by attending a lecture about balance. You cannot learn by watching an expert demonstration, no matter how well it's explained. You cannot learn by reading a detailed technical breakdown of the physics involved. You learn by getting on the bike, wobbling, falling off, getting back on, and wobbling less. The knowledge that enables you to ride is procedural — it lives in doing, not in understanding.
AI is the same. You can listen to the most compelling keynote about large language models, agentic workflows, and the future of finance automation. You can nod along, take notes, and feel genuinely energised. And then you can sit back down at your computer and not know how to start — because the gap between conceptual understanding and practical capability is exactly as wide after the conference as it was before.
The only thing that closes that gap is using it. Regularly. On real problems. With real stakes.
Nobody Goes to an Excel Conference
Think about how you learned Excel. I mean really think about it.
You almost certainly didn't attend a structured Excel learning program before you started using it at work. You probably sat down, opened a file, needed to do something, figured out roughly how to do it, got it wrong, fixed it, and gradually developed capability through accumulated problem-solving. Someone showed you VLOOKUP when you needed VLOOKUP. You learned pivot tables when you had data that needed pivoting. The learning was driven by problems, not by curricula.
Nobody holds an Excel Summit with $800 tickets. Nobody has a two-day immersive Excel experience to prepare you for "the Excel revolution." And yet finance teams in Australia and globally developed extraordinary Excel capability through nothing more sophisticated than: here is the tool, here is a real problem, figure it out.
AI is exactly the same. The tool is more powerful, the problems it can help with are broader, and the potential upside is significantly larger. But the learning mechanism is identical: get on the bike.
Finance Professionals Are Actually Wired for This
Here's what I didn't expect when I started using AI seriously: my finance background made me better at it, not worse.
Not better at prompting in any sophisticated technical sense. Better at knowing when the output was wrong. Better at having a prior expectation of what a correct answer should look like. Better at the instinctive audit that says "wait, this number doesn't feel right" — which is the same instinct that catches a balance sheet that doesn't balance, a variance explanation that doesn't explain anything, or a payroll calculation that's slightly too neat to be real.
When I was building my portfolio tracker last weekend and the Silver ETF suddenly made my portfolio look like $3 million, I didn't need to debug the code to know something was wrong. I just knew. Twenty years of looking at financial numbers means you develop a sense for them — a pattern recognition that kicks in before the analysis does. That instinct is exactly what you need when AI produces plausible-looking output that is nonetheless incorrect.
Most non-finance people who use AI for financial tasks don't have that. They see a number, it looks reasonable, they accept it. Finance professionals are trained sceptics. That's a genuine advantage in the AI era, not an obstacle to overcome.
Your Excel Spreadsheets Are the Starting Line
There's a transition that I think is about to happen in finance teams everywhere, and I want to name it clearly because I think it's the most practical entry point for people who want to actually use AI rather than just understand it.
Every finance team in Australia is running on Excel. Management reporting packs, reconciliation templates, budget models, cashflow forecasts, payroll calculators — Excel is the operating system of finance, and it has been for thirty years. And here's the thing: every single one of those spreadsheets is a candidate for a better version, built with vibe coding and AI, that does the same job faster, with less manual input, and with better error detection.
You don't need to know how to code. You don't need to understand programming languages. You need to understand what the spreadsheet is supposed to do — which you already do, because you built it or you use it every day. That domain knowledge is the most valuable input. The AI handles the rest.
I'd genuinely encourage finance professionals who enjoy Excel — who get satisfaction from a well-built model, a tight reconciliation, or a clear dashboard — to try this. Pick one spreadsheet. Describe to an AI what it does, what inputs it takes, what outputs it produces, and what you wish it could do better. See what comes back. The people who find this genuinely exciting will grow fast. The capability compounds.
Just as finance teams became the go-to for Excel expertise — colleagues would come to you with "can you build me a spreadsheet that does X?" — the same dynamic is coming for AI tools. The finance professionals who start now will be the ones colleagues come to in two years when they need something built.
This Isn't the Dot-Com Moment. It's the Spreadsheet Moment.
I want to end with a perspective that I hold with some conviction, even knowing it might be wrong.
There's a habit of comparing AI to previous waves of technology — dot-com, smartphones, cloud computing. The comparison is usually made to signal either excitement or caution: "this is as big as the internet" or "we've seen hype cycles before." I think both framings miss the point.
The comparison that feels right to me is further back. When computers arrived in finance departments in the 1980s, and then when spreadsheets followed, it wasn't an incremental improvement to existing work. It was a fundamental change in what a finance person could do with a day. Reconciliations that took a week took an afternoon. Models that required teams of people could be built by one person. The constraint changed from "how many people do you have?" to "how good is your spreadsheet?"
AI is doing that again. The constraint is shifting from "how much time does your team have?" to "how well can your team think, prompt, and verify?" That's a different kind of challenge — and a much better one, if you're positioned for it.
Conferences will help once you have enough experience to know what questions to ask. But experience comes first. Get on the bike.
At PFL, we've built our entire practice around the practical application of AI in finance — not the theory of it. If you want to understand what AI adoption actually looks like in a real finance function, and where it makes the most difference for organisations like yours, let's have that conversation.
Talk to PFL →- Professor Kim Dae-sik, KAIST — neuroscience perspective on skill acquisition and AI learning (referenced from public video commentary)
- CFO Connect — State of AI in Finance 2026: Adoption Trends, Tools, and CFO Roadmap
- Deloitte Australia — State of AI in the Enterprise 2026
Comments
Post a Comment