On the GMAT, Artificial Intelligence Wins

Artificial Intelligence:  The study and design of intelligent agents, in which an intelligent agent is a system that perceives its environment and takes actions that maximize its chances of success.

Talk to friends studying for the GMAT, or browse GMAT-related forums, and you’re bound to learn a bunch of if/then rules (many of which are helpful, some of which are counterproductive): “if a question asked uses the word ‘and’, you multiply; if it’s ‘or’, you add”; “if the verbs in a sentence are in different tenses, find the answer that puts them all in the same tense”; “if the word ‘being’ is in a sentence, eliminate it”; etc.

Watch a class go through a challenging problem and you’re bound to hear questions such as “I see that now, but how would I ever know to do that?” and “When a question uses (this word) can I always do that?”

And here’s the sum of most of those conversations: most GMAT students are trying to program themselves as test-takers.

Which is understandable – for most of your academic career you’ve been able – and encouraged – to do that.  See a quadratic equation?  Plug it into the quadratic formula.  What happened in 1492? Columbus sailed the ocean blue.  Need to write a persuasive essay?  Start with a thesis statement, then support paragraphs with examples for each…  You’ve been able to program yourself for academic success with if/then binary logic.  But if high school and even college were Web 2.0, business schools are looking for the future, and as we know from emerging technology the future is more than just binary programming. The future is Artificial Intelligence, which is a great metaphor for what you need in order to succeed on the GMAT.

Artificial Intelligence goes back to that definition – it’s not “if/then” binary statements, but rather an agent (you) that can take actions to maximize its chances of success.  It trains itself to recognize patterns and play probabilities, but to be able to reevaluate when one process looks unlikely to achieve its goal.  As a GMAT instructor, I’ve long felt that my role isn’t to “show you how to do this problem” but much more “to help show yourself how to do problems that you haven’t seen yet”.  Effective GMAT preparation, at least for those seeking the elite 700+ scores, is less about being programmed to solve known problems with known techniques, and much more about developing flexible techniques and mental agility.

Now, if this concept seems a bit vague, well, so is a lot of the GMAT.  And one form of GMAT AI involves taking abstract concepts and making them concrete with examples, so let’s do that.

Consider a few of the binary-program strategies that test-takers learn for Problem Solving questions:

Algebra

Backsolving

Factors, Multiples, and Number Properties

Many a GMAT examinee loves to be able to say “this is an algebra problem” and then begin to do algebra.  Or to identify “this is a good problem on which to plug in the answer choices.”  But take a look at this 700-level problem, which appears courtesy of the GMATPrep software (spoiler alert: you might see this on a practice test someday):

A boat traveled upstream a distance of 90 miles at an average speed of (v – 3) miles per hour and then traveled the same distance downstream at an average speed of (v + 3) miles per hour.  If the trip upstream took a half-hour longer than the trip downstream, how many hours did it take for the boat to travel downstream?

(A)  2.5

(B)  2.4

(C)  2.3

(D) 2.2

(E)  2.1

Now, consider a few elements in play here – the algebra looks like it will set up ugly, so you may want to backsolve.  But the given variable is v and the question asks you to solve for a time, so a quick backsolve isn’t really in the cards.  You’ll have to do some algebra.

Questions about this article? Email us or leave a comment below.