Back to Blog
Blog post image

AI might be revolutionary, but the masses won't move any time soon

May 10, 2023 8:33:58 AM

ChatGPT is going to take over the world and make you redundant within a year…

That's the fear many of you are battling with, isn't it?

Imagine a world where you just speak into your mic, and ChatGPT just gets it and builds a perfect model in minutes. It builds a model that's better than the one you would have built, it's as automated and user-friendly as it can possibly be, and it all just works.

As a modelling geek, I actually think this is pretty exciting! If you tell me I can have a robot do all the painfully repetitive build work for me then, honestly, just take my money now. I get to use my brain to work more on the project's scope and model spec, to manage and build relationships, and to invest more time in refining the key insights and outputs that I'll share with my stakeholders.

That's an argument that my business partner, Myles Arnott, wrote about recently here. It's not the main point I want to make in this blog though.

I'd like to argue that the majority of us Excel users are pretty safe for the foreseeable future, no matter how game-changing AI turns out to be

Even if ChatGPT or other AI tools offer revolutionary modelling solutions within a couple of years, I'd argue the majority of us are still safe for now (with a caveat or two). Here's why.

At Full Stack, without a doubt, the most impactful learning module we teach is Power Query. We've seen individuals and teams of finance professionals transform their working lives using Power Query. They've cut tasks that take multiple days to complete down to a few clicks of the Refresh button in Excel.

The crazy thing is that Power Query has been around for more than a decade! It's relatively easy to learn, and if you don't want to write a single line of code (like me) then you can still leverage Power Query to do some incredibly powerful ETL work. It's been there, waiting for finance professionals to pick it up and take advantage of for years. And yet, the majority of Excel users who would benefit the most just haven't made that leap.

My impression of the c750 million Excel users out there is that the vast majority are still struggling with the first few fundamental steps of best practice modelling and core functional skills. We still see thousands of modellers learning how to use SUMIFS, how to clearly and consistently structure their modelling logic, how to manage date logic, and how to use features like modelling flags properly.

In my opinion, even if the top 1% of modellers are doing crazy, groundbreaking work with AI, the masses will be many steps behind for years to come. That's not a criticism of the masses (far from it - I love helping the average Excel user take those first few training steps into what I'd call Intermediate skills territory). It's just my assessment of how these things tend to play out when the latest innovation hits the market.

Don't get too comfortable though. If your only notable skill is building good spreadsheets, then AI - and modelling tech in general - is definitely going to start eating away at your perceived value. You must be good at all of the other skills Myles mentions in his blog in the link above.

Why should anyone learn the fundamentals if AI can do all the work for us?

You could make the argument that inexperienced modellers don't need to know the basics of manual model development now that AI can do the heavy lifting. I actually think, if you're in your teens or early 20s and you're just starting out, there's a pretty strong case for this. You could, arguably, choose to become an expert in AI instead of becoming proficient in financial modelling.

My pushback against that argument is that, in almost any industry you can think of, it never hurts to understand the fundamentals; to know how all the underlying pieces fit together, and how you would build or operate a model (or whatever it is you work with) without the aid of advanced robots.

It would be like assuming today that there's no need to learn how to drive a car, because Teslas can do the driving for you. I'd still prefer to know that I, or the person at the wheel of the Tesla I'm sat in, knows what to do if there is a need for manual intervention. In addition, the reality is that most of us on the roads today aren't taking a mid-afternoon nap in a Tesla while it drives us from A to B.

When you learn how to drive a car, you learn all the basic techniques first: How to start on a hill, how to reverse park, how to do a 3-point turn. You also have to go through your driving theory assessment - learning the fundamental principles that could save your life before you're let loose on the road.

Do you understand exactly how a car works? Probably not. Do you need to? No (but it wouldn't hurt).

I love the idea of a future where we all jump in a Tesla, tell it where to take us, and we just sit back and let the car do all the hard work. But it's still enormously valuable to understand the fundamentals, at least until we're a few steps closer to that reality.

So should should you invest your time learning more about AI?

Definitely! I'm not trying to put you off signing up for the latest ChatGPT course. If you're interested in these new developments and you want to advance your skills and knowledge in this area, you should do so.

For the majority of Excel users who are probably still struggling with the basics, I'm just saying: don't panic. Your job is safe for now. Learn the fundamentals, and then learn the cutting-edge stuff.

And if you drive a Tesla and you're already just telling your car where to drive you… I bet my hill starts are better than yours 😉

Take our free 30-question skills assessment

See where you stand with your Excel, financial modelling, data visualisation & analytics, and modelling tech skills

Take Free Skills Assessment

Subscribe to our monthly modelling newsletter