Educators Should Think Like Hackers

If you’re with the times, you might have come across articles and opinions that heavily criticize the use of large language models (LLMs) among students at different levels. One such opinion piece is the New York magazine essay by James D. Walsh. He opens his essay by talking about Chungin “Roy” Lee, the now-suspended Columbia University student who created the interview coder software. This software is advertised as a tool that’ll help you cheat your way through FAANG level interviews. On its website, there are testimonials of people using it to answer various levels of LeetCode style questions, undetected. For those unfamiliar, LeetCode is a website that is used to practice interview-style questions, heavily focused on data structures and algorithms.

Walsh’s essay explored how university students are using LLMs to do things like essay writing, homework, and quizzes. It also talked about the panic among educators. Some have developed countermeasures that barely work, some have gone on to write essays that critique the current state of education, while some have decided to retire. One can conclude that it is not really the best time to be an educator, seeing as most students do not appreciate the value of their work.

The popular perspective is that students are no longer interested in putting effort into their learning and are happy to cheat their way through school. One question I hear from the older generation is, “Why would anyone spend so much money and time to get a university degree and not bother to engage with the material?”

To properly understand this dilemma, we need to take a few steps back. For as long as I can remember, a university degree has always been sold as a path towards social and class mobility. Most high-paying jobs would prefer candidates who have at least completed a bachelor’s degree. A lot of people go to university even when unsure about what to do with their lives. They know that they’re guaranteed something if they can coast through the higher education system. It doesn’t matter if their true passions are outside of their degree; they make it through the university because it gives them an edge.

The assumed primary aim of any formal education system is to increase the knowledge and understanding of its students. The system promises to teach them how to think critically and act intelligently in the world. Most people can agree that this is the target. I heard somewhere that higher education systems are in the business of preparing people for the future, which is mostly in line with the previous statements. Every system has methods and metrics to evaluate how well it has accomplished its goal. For a school system, the methods are through tests, exams, homework, and the metrics are CGPAs, letter grades, etc.

Although what has happened through the years is a manifestation of Goodhart’s law. I have written about this in a previous essay. “When a measure becomes a target, it ceases to be a good measure.” A student knows their degree and CGPA matter. Higher grades mean better CGPA, so any smart student is solving the optimization problem of maximizing their grades. Even when they might be investing time and effort to properly learn the material, they know that grades are the ultimate metric for success. It is not difficult to see why they resort to things that optimize this metric: memorization, cheating, plagiarism, etc. With easy access to LLMs, a student has the perfect tool to solve this optimization problem.

They can write multiple versions of an essay, cheat on homework and projects, within seconds. This is simply Goodhart’s law at play. The measure has become the target. The student sees the value in getting a degree, but they’re too focused on optimizing metrics even when no knowledge is retained.

I was the teaching assistant for a chemistry lab in my first year of graduate school. I opened one of the classes by telling the students, “None of this stuff really matters. It is the relationships you’d form by working with your team that counts”. It was a naive approach to tell the students that I cared more about their lab engagement than their grades. I realized later that a bunch of the kids were premed students who were purely interested in grade optimization. Some of these students would beg and argue over a 0.25 after they got their scores back.

I didn’t have much control over the method of evaluation, but I found it interesting how the students were under so much pressure to increase their scores, not really caring about the high-level content of the class. Still, I didn’t blame them. The underlying problem is the systems that make them act this way.

We shouldn’t blame and shame the students. Perhaps we are puzzled that people would self-elect themselves to get an education, but end up as grade optimizers. We fail to ask ourselves if our methods and approach are still relevant to the targets we intend. This is where the mindset of security and hacking becomes necessary. How do we adapt our evaluation methods to prevent malicious efforts?

One of the things that initially drew me into tech was the field of security. Everyone agrees that it is important, but somehow we’re always lacking. The field of security is highly adaptive, both on the offense and defense sides.

On the offense and defense side, there are two main players;

Black Hat Hackers: These are the bad people. They’re the ones who write exploits with the intent of breaching vulnerable systems for theft and manipulation. Under this category are the people who make the bad software and the people who use it. They’re the reason why you use a secure password and avoid clicking suspicious links.

White Hat Hackers: They’re the ones who test our systems to make sure they’re safe. They have the same set of skills as the black hat hackers, but they use them to keep systems safe. They simulate different methods to hack the systems they protect, and develop countermeasures when attacks are successful.

Both the Black and White Hat Hackers share the mindset of adaptability, understanding systems, identifying vulnerabilities, and creating countermeasures. In the world of security, if a method is revealed to be vulnerable, there’s usually a rally for better and more secure methods. Everyone in this field agrees that there’s nothing like a secure system. For every new “secure” method or system, countless people are working hard to find vulnerabilities and exploit them.

Just like the Black Hat Hacker will exploit systems for personal gain, the student who is looking for an easy way to optimize their grades will exploit the methods we use to evaluate them. They will find ways to “hack” the grading system, either by employing tools that make plagiarism detectors useless, leveraging AI to do their homework, or simply cheating in the old-fashioned way.

This is where the White Hat Educator comes in. Instead of complaining endlessly about the state of education, they try to develop new systems and methods to evaluate students. They know that if a machine can solve their entire exam or homework in a few minutes, then that method of evaluation is broken. They think about new ways to accomplish the original goal of education. They ask questions like: How do we make the student see the value in their education? How can we design evaluations that are cheat-proof and even better, don’t incentivize cheating? What metrics are important in the times we live in? What type of skills should we be testing for?

I find these questions to be very important. I was having lunch with a connection who is now a professor at Cornell, and he told me that most professors’ version of a “good grad student” is the type of grad student they were. This is relevant because we sometimes think that the metrics that made sense yesterday still apply today. Perhaps the definition of a “good student” 50 years ago is not the same in 2025. But the education system and the type of people it employs can be rigid.

The people who protect our digital systems know that the users might not act rationally. They account for the cases where the user doesn’t act in their own best interest. This is the nature of their job, and most do it diligently, not retreating to despair. If the modern student is not acting in their own best interest, we need to find ways to help them. While a student has more responsibility for their learning and education, the educator also has to help them learn and make sure that the systems they employ evolve and align with this goal.

Two methods that I think are effective and almost cheat-proof are presentations and projects. The way someone presents a topic and responds to their audience’s questions can be a good signal of how knowledgeable they are about said topic. In addition, executing a project requires multiple skills like planning, design, time management, etc. All of these can be aided by LLMs, but these methods require some human effort and mostly guarantee a unique outcome for every student. It’ll be rare for two students to present on something or execute a well-rounded project in the same way. While these methods are already in use, they mostly account for smaller portions of a students final grade. If we’re leaning towards methods that truly test a student’s ability, I think these avenues should be explored more.

Our methods and systems should constantly be revised to ensure that they work as intended. The modern educator, while unappreciated, must adapt to the changing landscape of available tools. We need White Hat Educators who help students see the beauty in their quest for knowledge.