Is Math Kind of Obsolete? Evaluation, Execution, and Mastering Mastery

I get a surprising range of responses when I tell people that I work as an “AI Transition Consultant.”

Maybe it shouldn’t be surprising. The advent of Artificial Intelligence is the kind of cultural sea-change that impacts all of us - it makes sense that everyone will have notes. Even so, I’m regularly taken aback by the variety, depth, and passion of opinion I hear from the people I meet who are nearly-universally motivated to give me their take on the topic whether they have any experience with the technology or not. 

It’s the kind of surprise I’ve come to delight in: no matter where I am or who I’m talking to I can always find a unique take. And frequently a challenging one.

“Math Is Kind of Obsolete.”

A perfect example of this reared its head in conversation recently with Greg, a new healthcare worker I met at my doctor’s office the other week. He was friendly, early in his career and in the throes of the novelty of his new role, and we traded small talk between blood-pressure tests and medication verifications that quickly turned into bigger talk when the subject of AI arose. He was curious about the kind of work that I do and more than happy to offer his own opinions with minimal encouragement.

“It’s pretty crazy,” he told me (a fairly common refrain), “how much it’s already changed. The technology didn’t even exist when I started high school and now it’s in everything. I use it a hundred times a day. When I was training there were so many things I was taught I was going to have to do every day that the computer has already taken over.”

“There’s stuff it can’t do, of course,” he was quick to point out, “but there’s so much that it can! And better and faster than I ever could. When you think about it, math is kind of obsolete!”

I don’t get a lot of opportunities to be aghast in day-to-day life, but here, to his delight, I had found one that left me with jaw hanging. Greg was a bit inexperienced, certainly, and perhaps overly starry-eyed about innovation - but clearly intelligent and apparently educated. He had to know that computing is made of math, right? And he worked in medicine: surely he must have been aware of the necessity of exactitude in the presumably-many evaluations and conversions and divisions required of him daily?

While his laughter at my responses suggested he was slightly exaggerating for effect, it was also clear that he was perfectly serious.

Removing the Minutiae

His point, once explained, wasn’t unreasoned or even entirely without merit.

Many of the calculations and measurements Greg would have once needed to do multiple times a day had already been subsumed by calculators or automated fields in his intake software well before he’d entered the domain. For him, AI promised to streamline things even further.

An AI could surface information from patient records of which he might not have immediate knowledge. It could suggest diagnostic pathways for him to explore by posing questions to his patients that he wouldn’t have considered. It could calculate dosage changes proactively from transcripts of conversations with the doctor.

Why should he care about the minutiae of any particular operation if the computer will do it anyway? Why even waste the time learning how to do it himself?

For him, the AI Revolution was removing the tedium so he could focus on the more fulfilling aspects of his work. 

Though somewhat disconcerted by his statement (and with the stern eyes of several former favored math teachers boring holes into the back of my conscience), I could hardly condemn such a sentiment.

A Most Discomfiting Error

Even several days later I found myself unable to leave what he had said alone.

I had to wonder a bit at why his remark had struck me so hard. I make something of a point not to shake my fist at “kids today” if I can help it; worrying that technology will rot our collective minds and destroy civilization is a tradition very likely as old as “kids today” themselves. And I can profess no particularly great personal love of numbers, even though I recognize they make up the foundation of my field. 

But the abject rejection and summary eulogization of the entire field of Mathematics stuck in my craw. 

What I eventually realized was that Greg’s statement was indicative of an error in underlying logic, and one that I see a lot of people at all career stages making as they learn how to use AI in their own work: it conflates an executive action with an evaluative one.

Execution Isn’t Evaluation

Even the simplest of tasks has both an evaluative component (What is the best way to secure these two boards?) and an executive component (Hammer this nail into that given spot). 

Tasks become more complex when more evaluative or executive components are added to them, while tools and technology alleviate complexity by removing or resolving them. 

The nature of work to this point in our history has necessitated compiling tasks into workflows to accomplish goals of ever-increasing complexity; manufacturing a jet engine involves a great many more steps than building a wooden fence, after all. In response, over time most organizations end up segmenting more and more of the executive and evaluative work in a bid to increase efficiency.

AI disrupts this by dynamically automating not just the executive components of a workflow but the simpler evaluative ones as well. The effect is less like adding a faster tool to the line and more like removing several of the line’s checkpoints at once. 

This is especially disruptive in organizations and roles into which evaluative work has been deliberately segmented; exactly the sort of work that HR has spent decades professionalizing. Screening, assessment, calibration, compliance, the endless triage of “Does this situation warrant escalation?” - these are evaluative tasks that were carved out and specialized precisely because judgment was the valuable resource.

This is the tension that Greg had picked up on: judgement has, to a greater extent than ever before, become automatable. 

You Can’t Evaluate What You Don’t Understand

The result is that work moves faster, but the evaluative tasks that remain are the hard ones. Every remaining checkpoint carries more weight because there are fewer of them, and because more work is arriving at each one.

Greg was right that the computer will do the calculation. He was wrong to conclude that the calculation was ever the point.

We don’t learn arithmetic so we can spend our lives doing long division by hand. We learn it so we know when the answer on the screen belongs in the world.

The automated field pulled the wrong value. The AI summarized the wrong patient history. The model confidently recommends something that doesn’t fit the context. None of those are calculations. None are failures of execution. They’re failures of evaluation. And you can’t evaluate what you don’t understand, because the error doesn’t announce itself; it arrives looking exactly like an answer.

That principle extends well beyond healthcare.

In HR, AI can screen résumés, draft performance summaries, identify themes in engagement surveys, and even suggest policy language. Those are remarkable capabilities. But they also make human judgment more valuable, not less. Someone still has to recognize when a promising candidate has been filtered out for the wrong reasons. Someone has to notice that the engagement data is telling two different stories depending on which employees you’re looking at. Someone has to ask whether the policy is technically compliant but culturally disastrous.

Those aren’t prompts. They’re judgments. And not automatable ones.

Judgement Takes a Larger Role

Here's the part I find most interesting, and I think it's where most of us actually live.

Work moves faster now. The tedious checkpoints are soon to be gone. But the judgment calls that are left are the hard ones, and there's ever more arriving at each one. The volume of judgment doesn't drop - its density increases. It concentrates.

For years, professional development often meant becoming more efficient at the work you already did. AI changes that equation. The most valuable skill isn’t memorizing every new tool or chasing every new model. It’s becoming exceptionally good at recognizing where your own understanding ends and then closing that gap before it becomes a blind spot.

So the person who gets ahead is the one whose judgment covers the most ground. Deep enough to evaluate well inside their own domain. Broad enough to evaluate at all in the domains that have quietly become theirs, because the work now moves fast enough to reach them.

The First Act of Judgement

And this is the thing almost none of us were taught: you can't go get the expertise you're missing if you don't know enough to notice it's missing.

Noticing is itself an evaluative act. It runs on the same foundations, and it comes first.  Before the analysis, before the decision, before you even know there's a question to ask.

The people who thrive here won't be the ones who know everything. They'll be the ones who reliably know what they don't know, and who have gotten fast at going and getting it.

That's a learnable skill. It's also, for most of us, an unpracticed one. Because for most of our careers the pace of work gave us the luxury of not needing it.

How to Build Better Judgement

Practically speaking, this is a different kind of reskilling than most of us are used to. Because it’s not about learning the next specific thing, it’s about learning how to learn. 

It’s about mastering mastery.

This is a lifelong endeavor, of course, but here are three things you can do now to take the first steps:

1. Get used to saying “I don’t know” People at every level are afraid to admit when they don't know something, but pretending at omniscience doesn't make anybody more effective. Practice admitting when you need to consult someone or something else before putting forward an answer, and work to build a culture where the people around you feel safe doing the same.

2. Confront your conclusions Probe the things you think you know, especially the ones outside your direct expertise. Audit a report you approve every week without really reading. Ask a colleague bluntly whether you're getting something wrong, and actually listen to the answer. Push back when somebody says something "everybody knows."

3. Cultivate curiosityThe first two habits work on the material already in front of you. This one is about going to find the material that isn't. Look for opportunities to expand into areas you haven’t considered before. Ask your IT staff what they've been reading lately. Sit in on another function's planning session. Join a class where you'll be the least experienced person in the discussion. Not just for the curriculum, but for the handful of other people whose gaps aren't the same as yours.

That last point is the whole reason Reframing HR built the HR Reframed Masterclass, a cohort masterclass for Senior HR and TA leaders. Not to teach prompts of the AI tool of the week, but to help leaders develop the mindset, judgement, and practical capabilities they’ll need as AI reshapes the work around them.

The Math Was Never the Point

I’ve thought about Greg’s comment a lot since that conversation. I hope he is right, and that AI clears away the tedium so he can spend his day on the work that actually matters to him.

But I want the person checking my dosage to have learned the math anyway. Not so he has to do it, necessarily, but so he can tell when it's wrong.

I think he was seeing something very real about the moment. AI is making certain types of skills dramatically less valuable than they once were. But it isn’t making expertise obsolete. It’s concentrating it. The execution is increasingly automated. The judgment isn’t.

And that’s a transition every one of us, and especially those of us responsible for developing the next iteration of Work, has to learn how to navigate.