Like any computer, Deep Thought was just following the directions. Unfortunately for the civilization - after 7.5 million years of waiting - the answer they got was absolutely meaningless. 

The Truth Is Out There

We're living in an age in which we have free access to leviathans of information. Google. LLMs. You're on Gregmat - a cornucopia of GRE prep information. But to get a good answer, we must ask the right question. With LLMs, a fancy name for this process is "prompt engineering". The LLM has a titanic amount of knowledge. If we can intelligently word the way we prompt it, we'll get more useful information.

For example, if I open up Chat GPT and ask it "how can I improve my life?", even if it knows many things about me from my previous chat history, its suggestions are likely to be pretty bland, or general. If improving my life is a topic I've thought about before, it's unlikely I'll think most - or any of its suggestions are useful. I might get frustrated and assume Chat GPT isn't worth my time.

However, if I design a more specific question, I might get a better answer. Let's say I want to improve my physical fitness. I could type in,

"how do I get in better shape"

or I could write a detailed prompt, like:

"Act as a certified personal trainer. I am a 40-year-old male, 5'11" and 210 lbs, working a sedentary office job. My primary goals are to lose 15 lbs over the next 4 months and increase my cardiovascular endurance so I don't feel winded walking up stairs. I haven't exercised consistently in over 5 years, but I have no major injuries, just occasional lower back stiffness. I can commit to exercising 3 days a week for about 45 minutes each session. I have access to basic home equipment: dumbbells (up to 25 lbs), resistance bands, and running shoes for outdoor walks/jogs. I dislike high-impact activities like jumping but enjoy walking and basic strength training. Please create a balanced, beginner-friendly 4-week introductory workout plan focusing on gradual progression for both cardio and strength, incorporating warm-ups and cool-downs. Also, provide 3 simple nutritional tips to support my weight loss goal without requiring a complex diet overhaul. Structure the plan week-by-week, outlining exercises, sets/reps, or duration for each workout day. Keep the tone encouraging but realistic."

Which prompt do you think will work better? The first step in getting a good answer means asking a good question.

In terms of GRE prep, it's understandable if your awareness of your problems is a bit hazy. There's a bunch of shit you need to do. Maybe your verbal score is 150 and you want it to be 160. Maybe your quant score is 155 and you want it to be 165. But to get help, you need to be willing to take the time to think about the specific components of your problem and focus on one of those components at a time - not just ask "how do I improve quant".

Sidebar: The "So you wanna get a 340" video series is a helpful way to define the different areas of prep and how we can work on them: i.e. things like foundation, strategies, untimed and time practice, etc.

Prompt Engineering for GRE Quant

For example, let's break down a common quant problem as stated by many people: "I understand the concepts, but I don't know how to apply them on more complex GRE questions".

We would probably look at a complex problem and ask things like:

• how quickly did you recognize the concepts involved?
• can you verbally rattle off the details of how any of these concepts work?
• are these concepts you have drilled extensively?
• did you notice which strategy(ies) can work on this problem?
• did you find a first step you can take on the question despite not yet knowing the "way to solve it"?

What we will probably uncover in this sort of line of questioning is that the problem isn't as simple as "I don't know how to apply concepts on more complex GRE questions". There may be multiple issues, which have to be addressed one at a time. This is great news for most people, since we now have actionable information, and we can form a plan to address these issues.

Now we can pose a better question, like "I was slow to recognize which concepts the question was testing - how can I get better at this?". Or, "I didn't realize I could choose numbers for this question - how can I build that skill?". Or, "It took me a long time to read and understand what the question was asking - how can I improve that?" Not only are questions like these probably going to get the person better answers from their tutor, from someone on the Gregmat forums, or from someone on Reddit -- but notice they are now questions that the person is more likely to be able to figure out the answers to on her own, since they're more bite-sized and defined.

Contrast this to advice this person might get if they claim their only problem is "I don't know how to apply them on more complex GRE questions" - someone might suggest "just practice, practice, practice", which probably will limit the progress they'll make on that particular issue, since merely practicing won't be targeting the specific weaknesses beneath the surface complaint.

Asking Good Questions For GRE Verbal

Of course, this works great for GRE verbal, too: instead of saying "I need help with Reading Comprehension questions", it will be much more useful to consider things like:

• what's this question asking you to do?
• what type of RC question is this?
• have you learned (and drilled) the strategy for this question type?
• what part of the passage is most useful here? How might we know that?
• where is the evidence for the wording in the right answer?
• why is a given choice wrong, exactly?
• what's the most attractive wrong answer, and why?
• what's unattractive about the right answer, and why?

As you can see, this analysis will help us identify discrete skills we can then work on with deliberate practice - skills that would have perhaps gone unnoticed if we just asked the person to practice more.

The Bottom Line

GRE mastery is an amalgam of lots and lots of little skills. Asking the right questions when navigating GRE prep means being more specific than broad. If we can first identify the precise nature of the problem, and then deliberately practice the skills involved, one at a time, we build up our GRE skill, brick by brick.