Symphony of Creative Destruction

If you're not cheating, you're not trying

This, but metaphorically.

Long before Silicon Valley started talking about disruption, there was a 19^th Century German fella who wrote about the creative destructive tendencies of capitalism. About a century later, an Austrian fella decided to put a bit more of a positive spin on it and the term “creative destruction” was born. To him this was a positive force, the way that the destructive forces of capitalism can, like a fire, clear away the old and create a new one.

We’ve seen this a lot over the last thirty years. Film cameras being replaced by digital cameras being replaced by smartphones. Letters replaced by email. Typewriters giving way to word processing software. The old keeps being burned down and replaced with the new. It, like change itself, is inevitable. We can embrace it or we can fight it. And although not all change is good, fighting change can often be worse.

Let’s pivot to crime and punishment, an issue I thought about greatly back when I thought I was going to be part of the criminal justice system. Why do people break the rules? There are a lot of theories on this, and, with all due respect to Cesare Beccaria, I’m going to do them all a great disservice by boiling this down to a few paragraphs. But I’m not discussing just penology¹ but all the ways we break the rules, whether the law, the rules of a game, or the rules at school or work.

One lesson from penology is that breaking the rules involves how the potential rule breaker views the punishment, particularly the likelihood and severity of punishment.² Societally we oft focus on severity. And although that has led to some incredibly ineffective policy decisions, it is an important factor. For example, let’s say the government can just monitor your vehicle and every time you speed, you get a speeding ticket. You’d probably be much less likely to speed, right? Afterall, part of why so many of you speed is because the likelihood of being caught is so low. But, what if the penalty for speeding was a fine of $0.01 cents per mile sped? At that point you’re probably going to sit down and study The Fast and the Furious movies for tips. Even certain punishment is meaningless without severity. Of course, if you flip the example and even with current speed detection techniques we made speeding a death penalty offense, you may be more likely to slow down. There’s a complex relationship between likelihood and severity, but they are the core elements of deterrence.

But they’re not alone. The punishment must be weighed against the benefits of breaking the rules. Or, more accurately, the perceived benefit. This is why to a drug addict it’s almost impossible to have a likelihood and severity high enough that it is worth more to them than getting high. And whether the benefit is merely perceived, or real, if it is greater than the perceived likelihood and severity of punishment, people break the rules. This is, for example, why in my adult life we’ve seen champions in the NFL, MLB, English Premier League, and Formula 1 all break the rules. The benefit to winning the championship is so much greater than any punishment they may later face that in a sporting context it is irresponsible not to cheat.³

You may be thinking to yourself “But wait, those teams you mentioned are all filthy cheaters!” And that’s because there’s an element of the penalty we didn’t discuss, the moral element. To some, the dishonor of being a cheater or a criminal is the heftiest punishment.

That desire to avoid being a cheat or a crook is what drives so many of us to not do things like cheat on our taxes or in our weekly card game. It’s also why so many of us speed. If a rule is rarely enforced, the moral and reputational stigma associated with it becomes lessened and it doesn’t feel like we’re doing anything wrong, even if we know it’s against the rules. The perfect example of this is academic cheating.

The research on this, even what I am about to link, seems weak. But combined with my anecdotal experience I feel comfortable saying that academic cheating is widespread. The profitability of paper mills certainly speaks to this. Although I, quite foolishly, never cheated for my own benefit,⁴ I still remember in elementary school passing around my test so my friends could copy it. Odds are every single one of you has, at some point, cheated in school. And it’s for the simple reason that you’re unlikely to get caught. If everyone is doing it, it’s not a crime. Academic cheating is a mess.

The good news is that it’s about to get messier thanks to our old friend ChatGPT. Just last week a bit of a furore began when a star gymnast at Louisiana State University endorsed what is, basically, an AI cheating tool. This is going on everywhere, as ChatGPT makes it incredibly simple to write short, competent essays and now schools are working hard to crackdown. And they shouldn’t.

Let’s start with this: banning something has a lifetime success rate of zero. Now, even imperfect bans are worthwhile. Murder has been with us since Cain, and yet, the societal interest in preventing it is so great an imperfect ban is still good. But we can acknowledge that banning ChatGPT written essays will not remove them entirely. So, let’s go back to our little crime and punishment matrix. The hope right now is that the likelihood of catching cheaters will be high.

OpenAI is working on a digital watermark. Someone has already created a tool that detects AI generated text with a fair degree of accuracy. And these will be useful right up to the point where these non-perfectly accurate tools unfairly punish a student whose parent is a lawyer. This early euphoria is, like all attempts to stymie academic cheating, likely to fail miserably.

Which means we need to look at our other elements. Penalties for academic cheating can be quite severe.⁵ But a large part of why people don’t cheat is, again, they don’t want to be a cheater. Which means the students self-deterring would be the most conscientious, the exact people society should want to reward.

Which means we must look to the final element: what is the perceived benefit of using ChatGPT to write essays for class? This is exactly why academic cheating is so rampant, we’ve placed great societal value on higher grades. I would argue that we should just reverse that, but I doubt many would agree. I also think restricting As to the type of exceptional work that an AI can’t do – yet – would be helpful, but turning the clock back on grade inflation is too much of a prisoner’s dilemma. We’re stuck with a high perceived value on using AI to cheat.

But what if there was another option?

One of the above linked articles includes a section that I believe deserves a closer look.

The New York City education department said on Thursday that it's blocking access to ChatGPT on school networks and devices over concerns about its "negative impacts on student learning, and concerns regarding the safety and accuracy of content."

To which I ask: what negative impact? How are students being negatively impacted by being allowed to use ChatGPT to write short essays? And this brings us back to our friends Marx and Schumpeter and creative destruction.

Back in my day, I always did excellently in college⁶ on assessments of my ability to remember facts that a few short years later I could find with a Google search. Perhaps there was a time when my ability to recall what year the Battle of Dien Bien Phu occurred was a valuable skill. But it sure wasn’t once I started carrying a repository of all human knowledge in my pocket. Which isn’t to say that knowing history isn’t valuable, it is, I wish more people knew it! It’s valuable to understand the significance of the Battle of Dien Bien Phu, but the trivia, which many of us grew up being tested on, is not.

This took 9.58 seconds.

Similarly, what is the value in being able to write a short, competent essay? Not much in a world where I can do this.

This took 41.23 seconds.

Much as penmanship was replaced with the need for strong typing skills or memorization replaced with the need for strong search skills, the ability to write a short essay is an outdated skill. So why not respond to the rise of AI enabled cheating by attacking the perceived benefit. Instead of a Quixotic quest to stop it, just destroy the perceived benefit by devaluing short essays in student assessment.

This is not a plan without drawbacks. I reached out to professors I know and asked why short essays are such a large part of their grading. Most of the answers fell along two lines. One is that students complain when required to write long papers.⁷ The other is that professors feel obligated to assign writing but are also so overworked that short essays are a nice compromise. Both are reasons I’m sympathetic to but don’t find compelling. However, one stated that he needs to see if students have engaged with and thought about an issue. And this is the most efficient way to do it. Which is a good reason for using them. Or, at least, it was. But now it is simply the folly of reconstructing a broken egg.

Undoubtedly, removing short essays from schoolwork is going to be hard. It was also hard for Kodak and Sears and James Cook to change the way they did things. So they didn’t. Of course, academia is not at the same risk. Creative destruction is a force of capitalism and as academia is not subject to market forces the way those businesses were. It could fight an endless, losing battle that benefits almost no one, except perhaps the most brazen and effective cheaters. And do we really need more future Wells Fargo executives?

But there are so many possible ways to assess students – in class exams, long papers, presentations, oral exams, creative projects – all of which seem more well correlated to the skills needed in the 21^st Century than writing short essays.

There’s a famous quote by John F. Kennedy that goes as follows:

The Chinese use two brush strokes to write the word 'crisis'. One brush stroke stands for danger; the other for opportunity. In a crisis, be aware of the danger–but recognize the opportunity.

Which is where academia is now. It can choose to react to ChatGPT in one of two ways. The first is what we’ve discussed, viewing it as a danger. The second is to view it as an opportunity. To update and change how it assesses students to something matching the modern world.⁸ And it seems that the latter, although hard, is the only path that could be fruitful. It’s time to allow the gale of creative destruction to revolutionize it from within, destroying the old and creating a new.

1. The study of the punishment of crime. But obviously I used this word because it makes me giggle.

2. Beccaria included the element of celerity, which in addition to referring to the swiftness of the punishment, also sounds like celery. We’ve got a lot of funny words this week people.

3. I picked examples of teams either being lightly fined or escaping punishment on a procedural issue. Even the severely punished teams probably don’t regret it.

4. This, in fact, is one of the most consequential moments in my life when I chose not to cheat on my Criminal Procedure exam and suffered greatly for it.

5. Although my panel of professors disagreed on this, with some telling me stories of students escaping quite lightly for what seems like serious offenses.

6. Well, almost always. I’ll see you in hell, unnamed Modern German History professor.

7. The answer I laughed at the most came from a professor whose identity I will guard to my dying day. “They also absolutely suck at writing long papers. They generally don’t have much going on in their brains, so when asked to write a long paper they mostly just repeat themselves and try to take up space.”

8. Is the entire system of higher education in need of a total reboot to match the Digital Age? Stay tuned!