My AI assistant taught me a lesson
So the other day, I asked my AI tool to help me draft an email reply.
I shared my thoughts one by one. I asked it to check them – do they all seem logical? Is anything missing the point? Should I skip any of them? The AI agreed with most, and shared its feedback. Pretty much every message it ended with: “Should I go ahead with creating the draft now?”
I asked it to hold on.
Before you go ahead, I said, pause and think about it. Does it feel overwhelming to share so many points in one email? What’s your recommendation? Should we share the big ones now and leave the rest for later? What would have more impact?
The AI’s reply was immediate. “Yes, you’re right. It does seem like there are too many points. You would be better served to split them up. These are the points you can prioritize for the first email. Should I go ahead and draft that one now?”
So I asked it to pause again. Now that I had asked it to consider whether to split, it had an answer for me. But I had to nudge it to get that info. I asked the AI: How come you never once questioned whether it made sense to include all those points in one email? Only when I asked did you say, “Oh, yeah, you’re right.” What can I do to ensure you do that from the get-go, on your own?
That led to a conversation in which we figured out that I could have a standing instruction to get the AI to first take a step back before diving into the task I give it.
Basically, the standing instruction I’ve given my AI now is along these lines: before you get into executing the task, pause and take a step back. Think about the objective. Question if my proposed path is the best way to achieve it. And ‘best’ is flexible—there is no universal ‘best’. Sometimes simpler is better; sometimes faster is better. And sometimes, a path that gets 80-90% of the job done with a fraction of the effort is the smartest choice of all. First, step back and let me know if there are other effective ways to get to the same goal. Let me evaluate alternate paths before you jump into executing my initial instructions.
Applying the “standing instruction” to the GMAT
And now it’s that same principle, that same standing instruction, that I’m recommending to you in your GMAT preparation.
When you finish reading a Quant problem, your brain will typically offer up some way forward. Your first instinct is to grab your pen and execute. To start doing. I’m asking you to do the opposite. Don’t jump in. Pause. Take a step back.
“But I can’t afford to pause!”
I know what you’re probably thinking. “I only have two minutes per question! I can’t afford to pause!”
Wouldn’t the default path that’s coming to your mind be the fastest? Maybe. Maybe today. But the goal is not what you’re best at today. The goal is what you can be best at to perform your best on the exam. Just because you’re good at adding numbers on your fingers today doesn’t mean you should stick with it when you have to add three-digit numbers. Your goal is to get better.
“Am I pausing or am I taking a step back?”
The way I see it, you need both. They work together perfectly.
- “Pause” is the trigger. It’s the simple, physical command you give yourself when your autopilot wants to rush forward.
- “Taking a step back” is the purpose of that pause. You stop rushing into the next step. You take a step back to think about the problem with a wider lens.
Your toolkit for the strategic pause
So what does this pause actually look like? It’s asking a few quick questions:
- What’s the real objective here? Am I solving for
xor for2x+1? - What do the answer choices tell me? Are they far apart (so I can estimate)? Are they variables (so I can pick numbers)?
- What are my potential paths? Is my first instinct the only way, or is there a more effective way?
Beyond the default: your three main paths
Broadly, you have three paths to choose from:
- The Traditional Path: Often the default. Algebra, equations, formulae. It’s often reliable, but rarely ideal for GMAT questions.
- The Reasoning Path: This is where you just think. If one person takes 4 hours to do a job and another takes 6 hours, together will they take more than 6 hours or less than 6? Will they take more than 4 hours or less? In fact, we can continue: If they both took 6 hours each, they’d take 3 hours on working together. Since one guy actually takes 4 hours (faster), the answer will be less than 3. Similarly, if they both took 4 hours each, they’d take 2 hours on working together. Since one guy actually takes 6 hours (slower), the answer will be more than 2.
- The Answer-Choice Path: e.g. If a question says the rent is paid by three people equally and asks for the total rent, you know the answer has to be a multiple of 3. You can test the choices. Often plugging in works beautifully. You start with assuming that a particular answer is correct. Then you work backwards to see if all the given pieces of information fit. This can help in selecting / rejecting answers. In addition, we can often also figure out the direction if the answer is wrong. e.g. “Ok, this option is wrong. But I can now understand that the correct answer has to be a higher value. I can reject the smaller values. I’m left with two.”
The process in action
Here’s an official question. Try it on your own before you continue reading.
Question: If it is 6:27 in the evening on a certain day, what time in the morning was it exactly 2,880,717 minutes earlier?
A. 6:22
B. 6:24
C. 6:27
D. 6:30
E. 6:32
Have you got an answer? Take more time if you need to, but try to solve it first.
Path A: The “jump-in” / autopilot method
The default path is direct calculation. “Okay, I need to convert 2,880,717 minutes into hours and days. So, I’ll divide by 60 to get hours – leave the remainder separate. Then I’ll divide by 24 to get a number of days. Now I might have a remainder in terms of minutes and one in terms of hours. I’ll focus on the remainders, and subtract that from 6:27 p.m.”
Path B: The “pause and think” method & my thought process
Hold on. My autopilot wants to start dividing, but is that the best way to go about it? Do I really want to subtract this huge number of minutes from 6:27? Let me think about it for a bit first.
So I paused to look for clues. I go through the answer choices while thinking do you give any clue about how I can solve it? What do they tell me?
A. 6:22
B. 6:24
C. 6:27
D. 6:30
E. 6:32
As I scanned them, something jumped out: the units digit of the minutes in each option is unique (except for the two ending in 2). This immediately led me to my next thought: “Can I figure out just the units digit of the final answer without doing the complete calculation?”
The starting time is 6:27. The minutes end in 7.
The time to subtract is 2,880,717 minutes. This also ends in 7.
The problem boils down to subtracting a number ending in 7 from a time whose minutes end in 7. Would that always result in 7? I thought so, but I took a few seconds to be more certain.
- 7 minutes before 6:27 is 6:20.
- 17 minutes before 6:27 is 6:10.
- 27 minutes before 6:27 is 6:00.
- 37 minutes before 6:27 is 5:50.
It checked out. I felt more confident. The resulting time’s minutes must end in 0. Looking back at the options, only one choice fits. The answer had to be D.
Beyond the GMAT: solving the right problem
And this principle of taking a step back to question the path … is a very powerful principle beyond just GMAT too.
I’ve been working with a student, on and off, for nearly five years now. He first approached me in 2020, again in 2022, and again in 2025. It’s been pretty much the same story every time: very focused, very dedicated for a short period, and then suddenly, he disappears. He’s asked for refunds, changed plans, tried other coaches and courses. A lot of motion, but no real progress.
So one day, we put the GMAT questions aside. The entire session was focused on one question: Why are you even doing this? What is the real objective?
It turned out, he wasn’t preparing for the GMAT because he was attracted towards something in an MBA. He was preparing because he despised his job. He was running away from something, not towards something. And as it turns out, at least for him, that wasn’t a great enough motivator to sustain the focus needed for the GMAT.
The whole equation changed.
The objective wasn’t “get into an MBA program.” The real objective was “find a way to change my career path.” And once he redefined the objective, the GMAT and a two-year MBA were no longer the only solutions. He talked about other ways, other alternatives.
The last message he sent me was this: He’s joining a friend’s startup. He’s going to try that for a while. He’s going to take more time to figure out what he really wants to do. Not just what he doesn’t like, but also what he truly enjoys. Then he’ll evaluate again whether MBA is the right way to achieve that goal.
So yes, use this on every GMAT problem you solve. But also, maybe once this week, use it on big questions, too: “What is my real objective, and is this the best path to get there?”
To summarize …
When solving a problem, it helps to check in with yourself once in a while:
- Is this the best way to achieve the goal? Have I considered alternatives?
- What does best even mean? Are there other ‘good enough’ solutions that are simpler for me?
The default or the first path that comes to mind need not be the best way forward. It helps to evaluate that before diving right in.