Is mathematical reasoning still worth the effort now that artificial intelligence is helping crack problems that left mathematicians stuck for decades? The question sounds like a provocation, but serious people are thinking about it. One of the most interesting answers came from someone who lived through this shift inside his own research.
In an interview with MIT News, published on November 17, 2026, mathematician John Urschel talked about what it was like to work with AI in his studies for the first time. His view is not against the technology. It is not starry-eyed either. For Urschel, the machine may hand you the answer, but the real value lies in understanding how you got there.
In the next few sections, you will see what he argues, why this idea carries so much weight for math in the age of AI, and how it connects to logical thinking in marketing, in the daily work of people who deal with campaigns, data, and digital products in Brazil. Spoiler: reading a performance report has more in common with solving a puzzle than you might think.
Who Is John Urschel
According to MIT News, Urschel is an assistant professor in the MIT Department of Mathematics, the same school where he earned his PhD. So far, a solid academic résumé. The interesting detail comes before that.
Before finishing his doctorate, he played football as an offensive lineman for the Baltimore Ravens. In plain terms, he was an athlete who trained for strength so he could hold the line on the field.
That background helps explain the metaphor he uses. The training is different now. Instead of physical muscle, he wants to build up the critical thinking of his students. And just like at any gym, nobody gets stronger by watching someone else lift weights.
In the classroom, the report says, he has two goals. The first is to help students train their problem-solving skills and their ability to think quantitatively. The second depends on the course: passing on the big ideas, tools, and techniques of each subject, so students can use that knowledge when they face real problems out in the world.
When AI Entered His Research
According to MIT News, Urschel recently posted two preprints on arXiv. One deals with the growth factor of random matrices. The other looks at worst-case behavior in Gaussian elimination, a classic method for solving systems of equations.
In both papers, he combined human insight with AI tools to tackle problems that, according to the outlet, had challenged mathematicians for decades. These were the first papers in which he used the machine, and he admits the experience feels different.
His comparison is very down to earth. Working with AI is like having a research colleague, only this colleague has what seem like unlimited resources and does certain tasks much faster than any person could. Just as computers already do arithmetic better than we do, he acknowledges that these systems can now handle some kinds of math far more efficiently.
Notice how careful that wording is. He talks about certain kinds of math, not all of math. That nuance gets lost easily once the topic hits the headlines, and it is exactly what holds up the rest of the conversation.
Why Mathematical Reasoning Still Matters
If the machine is so fast, why bother? The answer from Urschel comes down to one word: understanding. Getting a ready-made answer is not the same as grasping what is going on.
He explains that when an AI delivers the proof of a mathematical fact, the real value of that proof shows up when a human being absorbs how the machine put the ideas together. Then that person needs to internalize the reasoning and be able to pass it on. Without that step, the proof is just a nice-looking result on a screen.
That is why, according to the interview, he handles students in different ways. With those who use AI, the focus is always on understanding. With those who do not, the focus is on building problem-solving skills, something math has been doing for centuries.
His reasoning wraps up in a simple way: as long as there is room for humans to think and try to solve problems, and he believes there will be, training those skills is still worth it.
The Idea of Productive Struggle
One of his ideas sums up the interview nicely. Broadly speaking, Urschel says people do not like the word struggle, but when we do not wrestle with material on our own, we risk feeling like we learned something without actually learning it.
Anyone who has watched an entire tutorial, found it crystal clear, and then froze when trying it alone knows exactly what he means. That feeling of mastery can fool you.
Take an illustrative example: someone who wants to learn how to build spreadsheet formulas. If they just copy the formula the AI suggested, the sheet works that day. The next week, when the data layout changes a little, they have no idea where to start. Now picture the same person racking their brain for a few minutes before asking for help. They make mistakes, test things, and figure out the logic. The next time something changes, they fix it themselves.
MIT News also points out that a report from an MIT committee focused on the use of AI in teaching, learning, and research training highlighted the importance of this necessary, productive effort, along with the collaborative side of learning. So Urschel is not alone in this view.
That shared side also shows up in the article, which mentions the value of in-person education on campus, where students go through the struggle of solving problems and the connection with other people. Learning together, debating, making mistakes in front of someone. All of that is part of the training.
The Other Side: AI as a Tool That Levels the Playing Field
It would be easy to paint Urschel as a critic of technology. But he is not. In the interview, he says the rise of AI should be seen as a huge win for education.
The reason is simple. An easy-to-access technology that opens the door to a huge amount of knowledge helps level the playing field for people who never had that access before. He even mentions that neither of his parents was a teacher, which gives the argument a personal touch. This point ties in closely with what we have already covered about the new era of training and upskilling with artificial intelligence.
For a Brazilian audience, this part makes a lot of sense. Many people learn on their own, after work, with no teacher around to answer questions. Having a tool that explains a concept five different ways, at any hour, is a real gain.
That is where the balance lies. From a reading of the interview, AI widens the reach and speeds up the journey, but the thinking still belongs to the student. And he makes a personal confession that says a lot: for him, the joy of solving a problem is still greater when he does it alone, without the machine.
Logical Thinking in Marketing: What It Changes for You
Now let us bring the conversation closer to home. People who work in digital marketing are surrounded by dashboards, automated reports, and AI-generated summaries. They are a huge help, no doubt. But the risk Urschel points to in math shows up here in exactly the same way: thinking you understand the numbers just because you read the conclusion from the tool.
Logical thinking in marketing starts with uncomfortable questions. Did that conversion really come from that ad, or is the attribution model giving all the credit to the last click? Did the A/B test run long enough? Is the metric that went up actually tied to the goal of the client, or does it just look good on a slide?
None of these questions call for advanced math. They call for attention.
Using AI as an analysis partner makes perfect sense, and there are plenty of ways to use artificial intelligence in digital marketing that save hours of work. What sets you apart is being able to explain, in your own words, to your team or your client, why that number makes sense.
An Everyday Example
Picture a small agency working with an online clothing store. The AI tool generates the monthly report and flags a remarketing campaign as the big star. Everything is ready to send to the client.
Before hitting send, the analyst decides to take a closer look. She notices the period lined up with a big sale on the site, and that many of the people reached were already close to buying. The campaign helped, but maybe not as much as the summary suggested.
That check does not take a complicated formula. It takes curiosity and the habit of being suspicious of easy answers, which is exactly the muscle Urschel wants to see trained.
The same goes for people working on digital products. Imagine, for example, a team that asks AI to summarize hundreds of user comments. The summary says the main problem is price. But after reading a sample carefully, someone notices that a lot of people complain about price right after getting stuck on a confusing sign-up step. The underlying problem may be the experience, not the cost. Without that human reading, the team would fix the wrong thing.
How to Keep Your Math Muscle in Shape
The good news is that nobody has to become a mathematician to train this kind of reasoning. You can build a few habits into your routine without giving up your tools:
- Try to solve part of the problem on your own before asking AI for help, even if it is just a rough hypothesis.
- Ask the tool to show its reasoning step by step, not just the final result.
- Rewrite what the AI produced in your own words, as if you were explaining it to a new coworker.
- Set aside a little time for simple logic challenges, like sudoku or crosswords.
That last item has everything to do with Urschel himself. He says that ever since he was very young, he loved puzzles of all kinds, from sudoku to crosswords to the math puzzles of Martin Gardner. For him, the fun is in facing something you do not understand yet, slowly getting your bearings, and working your way toward understanding.
The third habit, by the way, may be the most powerful one. When you try to explain something and get stuck halfway, you find out right away what you still do not understand. It is an honest, quick, and free test.
Notice that none of these habits tell you to drop AI. The idea is to use the tool as a coach, not as someone who lifts the weight for you.
Those Who Understand Stay Ahead
The message from John Urschel is clear: AI speeds things up, cuts corners, and has already helped tackle old problems, but you are the one who needs to understand. Whether in math or in a campaign report, the value lies in absorbing the logic and being able to explain it.
In this landscape, the people who gain ground are those who think alongside the machine, question it, adjust, and own the analysis, not those who simply pass along whatever it delivers. That is the heart of math in the age of AI, and it also applies to anyone working with artificial intelligence and automation in digital marketing.
The next time a report lands on your desk ready to go, take a few minutes to rework the reasoning yourself before accepting the conclusion. It is a small workout. But it is what keeps your mind sharp.
