Lance Armstrong was a national hero. The cancer-surviving, seven-time Tour de France winner was one of the most recognizable figures in America at the height of his powers. And then it all came crashing down when a federal investigation revealed that Armstrong had taken performance-enhancing drugs. Overnight, his sponsorships disappeared, and he was stripped of his titles- a hero to pariah tale.
Armstrong faced perverse incentives. Having dedicated his life to cycling, battling back from devastating disease, he was left with a difficult choice: Stay clean and be uncompetitive at the highest level, or cheat and give yourself a chance.1 He decided to dope, and despite his amazing accomplishments, Armstrong today is remembered as a cheat rather than someone who achieved the nigh impossible and inspired millions.
This cautionary tale illustrates the moral quandary when one has to choose between cheating or failure. And I see these perverse incentives growing in the education sphere with ChatGPT and other LLMs upending our traditional school model. Article after article is being written about students cheating with these new tools- the most commonly cited statistic claims 90% of students used an LLM to help with homework. A recurrent theme in these articles is blaming the students, accusing them of laziness and poor moral judgment. But I see instead a system-wide failure that reeks of moral hazard to the point of entrapment. The failure here is not the student’s moral judgment, but instead the current incentive structure of our school system that prioritizes performance over learning.
The modern era of American education only dates back to the 1940s and the GI Bill. At the turn of the 20th century, only ~5% of Americans attended college, and most colleges operated under a standard exam system. If you passed the subject tests, you were admitted- albeit as long as you were the right race, gender or religion.2 This system was in its way objective. Students were not competing with one another for admission; they were competing against a set standard of knowledge. The G.I Bill changed this dynamic by rapidly democratizing college attendance- today over 63% of Americans over 18 have some post-high school education. This rapid increase in students strained the traditional University admission process. Prestigious schools now had to make a choice- lower their admittance rates or expand their student bodies. Not surprisingly, all chose to lower their admittance rates, further solidifying their role as status signalers in the modern age of intelligence.
By embracing their roles as status signalers, universities traded an objective standard for admission for a relative standard- pitting applicants against one another in a zero-sum race. This had significant knock-down effects in American society. Parents now had to focus significantly more attention and resources on their child’s development or risk other children passing them by. Free range childhoods became a thing of the past. Now, stereotypes of tiger moms and over-involved baseball dads abound as parents placed their children into more and more scheduled activities. All mostly driven by the dream of getting their child into the right college (or getting it paid for).
This combination of parental pressure and high societal stakes places a heavy burden on the modern high schooler. A seminal 2003 study revealed that teenagers expected to attend college (predominantly affluent teens) were three times more likely to be clinically depressed or abuse illegal drugs.3 The authors of the study hypothesized this illegal drug use was self-medication for the intense societal pressure and resulting anxiety that comes from status competition.
Now, let’s layer on top of this pressure the emergence of LLMs. Today’s high schooler now has access at the click of a button to a nearly all-knowing guide, and the temptation to replace their own work with the work of an LLM is immense. They are trapped in a classic Prisoner’s Dilemma, just like Lance Armstrong. Anyone who uses an LLM is significantly likelier to outperform their peers on grades. Therefore, every individual’s incentive is to use an LLM to replace their work, or risk falling behind their peer group. I have seen these perverse incentives firsthand with two students that I have discussed this issue with- both expressed frustration that peers who outsourced their work to an LLM received better grades
In response, most schools have returned to the worn playbook of draconian threats and severe punishment- a strategy I see as both ineffective and damaging. One of the students I chatted with described a witch-hunt culture where students are accused without evidence of LLM use, and they are considered guilty until they can prove their own innocence with pre-work. Not only is this a basic violation of liberal values, but as LLMs improve, they become increasingly indistinguishable from actual human work, and cheaters will be smart enough to create the evidence needed to pass off their work. Therefore, such a culture will only discourage high-performing kids in the long run- if I am going to be accused of cheating without cause, I may as well do it and reap the benefit- a grim, but logical line of reasoning. In addition, the proliferation of LLMs into the working world suggests that interacting with LLMs is an important skill for most jobs in the future. If our school systems do not embrace LLM use as a part of a student’s curriculum, they will quickly become outdated at the current rate of change.
Fortunately, there are feasible solutions to the challenges LLMs pose in education. The UK university system uses a single, in-class test at the end of a learning period to determine grades. The Oxford/Cambridge system of oral tutoring sessions also seems robust to this changing learning environment. Even the Socratic method of the Ancient Greeks feels well positioned for the LLM age. However, I propose something a bit deeper to consider.
I think the issue with US education lies at the core mission of the grading system- identifying the highest caliber of intellectual and driven kids and sorting them into high-performing schools to signal status. Grading served the US extremely well over the last century, as the value of human intelligence increased exponentially. However, LLMs completely turn the table on the importance of intelligence. Now, every American has access to a super intelligence that is becoming smarter every day. In this perspective, human intelligence is now a commodity whose value will decline over time, and our education system will similarly decline in value if it continues prioritizing identifying intelligence.
So, what should replace identifying intelligence as a core value of the education system in the Age of AI? I propose curiosity and creativity. We are entering an absolute golden age of self-development- where humans increasingly have all the information in the world at their fingertips. Everyone now has access to an elite tutor, editor or coach. I think the best comparison is the world of chess. Today’s chess players are significantly better than their predecessors- the top 45 players today are ranked higher than the best player from 1981.4 This is largely thanks to the help of computers who have given human players better opponents to practice against and learn from. In the same way, the LLMs of tomorrow could serve as the ultimate whetstones, sharpening human intellect with none of the traditional constraints of time or resources. However, to take full advantage of this gift, we need to teach our children to love to learn.
Loving to learn is not fostered in an environment where every test is life or death. Or when beating a classmate is the ultimate goal. Or when education outcomes equal status. Loving to learn is fostered by appealing to children’s individual interests. By creating a cycle where failure is not to be avoided, but encouraged, as it is a pivotal piece of a life-long journey.
I will leave you with a poem that I think is surprisingly relevant to this debate, Ithaka by CF Cavafy. A retelling of the story of Odysseus, it highlights a truth that Lance Armstrong forgot along the way. Ultimately, it is the experiences and the journey that define our life, not the outcome, and any shortcuts come at the expense of our own growth. Today, the incentives of the current education system promote shortcuts left and right to achieve status, but I am quite certain the Age of AI will belong to the learners rather than the A students.
Ithaka
By C. P. Cavafy
As you set out for Ithaka
hope your road is a long one,
full of adventure, full of discovery.
Laistrygonians, Cyclops,
angry Poseidon—don’t be afraid of them:
you’ll never find things like that on your way
as long as you keep your thoughts raised high,
as long as a rare excitement
stirs your spirit and your body.
Laistrygonians, Cyclops,
wild Poseidon—you won’t encounter them
unless you bring them along inside your soul,
unless your soul sets them up in front of you.
Hope your road is a long one.
May there be many summer mornings when,
with what pleasure, what joy,
you enter harbors you’re seeing for the first time;
may you stop at Phoenician trading stations
to buy fine things,
mother of pearl and coral, amber and ebony,
sensual perfume of every kind—
as many sensual perfumes as you can;
and may you visit many Egyptian cities
to learn and go on learning from their scholars.
Keep Ithaka always in your mind.
Arriving there is what you’re destined for.
But don’t hurry the journey at all.
Better if it lasts for years,
so you’re old by the time you reach the island,
wealthy with all you’ve gained on the way,
not expecting Ithaka to make you rich.
Ithaka gave you the marvelous journey.
Without her you wouldn't have set out.
She has nothing left to give you now.
And if you find her poor, Ithaka won’t have fooled you.
Wise as you will have become, so full of experience,
you’ll have understood by then what these Ithakas mean.
20 of 21 podium finishers in his era that tested positive or admitting to doping
Statistics: Education in America, 1860-1950 | Gilder Lehrman Institute of American History
(PDF) Children of the Affluent Challenges to Well-Being
2011-11_ACG13_Understanding_Performance.pdf
Another heater