A recurring theme I have observed in rationalism-type communities is a sort of moralistic fatalism or nihilism in regard to AI cheating, particularly in the context of the classroom or college, which is that cheating is good or justified, and futile or hopeless to stop. Colleges must accept that cheating is the norm, or that students are justified in cheating. Others defend Chat GPT and related programs as aiding in learning, so it’s not cheating if it helps students learn better.
A month ago, as typified by this sentiment, Zvi wrote a huge article blaming boring coursework as giving moral cover for cheating, and concluding that cheating may be a ‘net good’ by destroying credentialism and improving learning:
Ultimately, this particular cataclysm is not one I am so worried about. I don’t think our children were learning before, and they have much better opportunity to do so now. I don’t think they were acting with or being selected for integrity at university before, either. And if this destroys the value of degrees? Mostly, I’d say: Good.
And that detection is futile:
Trying to ‘catch’ the ‘cheating’ is pointless. It won’t work. Trying only turns this at best into a battle over obscuring tool use and makes the whole experience adversarial.
Colleges must adapt to the use of AI assistants, and if in the process this destroys the integrity of institution itself, so be it. You got to break eggs to make omelets. This is not only nihilistic, but also echoes similarities to the same reasoning the left invokes when justifying social unrest or de-policing, “If people were not being oppressed they would not have to resort to stealing.”
All of this comes of as morally inconsistent to me, from people who otherwise hold consistency as a virtue. Is interesting how many rationalist/centrist types are confident in the power of deterrence (e.g. policing and prisons) to stop crime, yet accept a fatalistic or defeatist attitude when it comes stopping AI cheating, or justifying said cheating or blaming the colleges.
I wonder why this is. It’s possible that the type of people who commit crimes are less deserving of sympathy, or are more distant tribally (low-income, uneducated criminals compared to high-SES, high-IQ students). Or that cheating in college is victimless, unlike actual crime. The ‘victim’ is the institution of credentialism, not an individual or a tangible loss with a monetary value. It’s harder to summon sympathy for employers, teachers, or administrators.
But it’s not entirely victimless in the sense that employers pay a large wage premium on the assumption that graduates are more competent, not that they are better cheaters. This presumably could have secondary economic effects such as loss of productivity and lowered living standards. Cheating also devalues degrees, including from graduates who didn’t cheat, as it’s now assumed that they did. This creates an incentive for everyone to cheat; what is there to lose when that bridge has been crossed and there is no turning back? It also means wasted money on tuition for a credential that is devalued.
Of course, Zvi and others can argue that this is ‘good and justified’, and will lead to much needed reforms of education and less reliance on degrees for hiring, but let’s not pretend the left doesn’t make similar arguments as well.
Moreover, I disagree about the defeatism. If prisons and police work as deterrence and crime mitigators, why can’t this be applied to cheating? Why would negative incentives work in the context of ‘law and order’ and not academia? People get mad when you suggest that cheating can be stopped or if you take a zero-tolerance approach to it. But this is no different from tough policing, which many of those same people agree is effective.
Why does the psychology of deterrence work for policing and not cheating? Just make the downside of cheating bad enough (e.g. automatic ‘zero’ in the course or assignment, or even expulsion) to dissuade enough people from doing it. As Zvi notes, cheating is hard to detect, but this does not imply trying to piggyback a moral justification for it. Just as the vast majority of shoplifters are not caught and that stopping shoplifting is difficult, it’s not as if society as a whole has suddenly condoned it.