The cheating cycle in academics is kind of interesting. A few years ago when getting into programming I took an assembly language course, basically a feeder pr

The cheating cycle in academics is kind of interesting. A few years ago when getting into programming I took an assembly language course, basically a feeder program at a community college for students transferring to 4-year CS programs. The final project was some ridiculously complicated program written in MIPS assembly with lots of functions and custom system calls, that you were supposed to debug and add to. Hand-editing saved registers on the stack and so on. I didn't care at all about the grade, as I was just there to learn the material (and was irritated about learning MIPS instead of something useful like RISC-V), but I tried working through it for a while, an incredibly tedious exercise. Then I talked to a few other students, and literally every single one got the code off some online site. It was auto-graded via online submission, so they got their good grade and moved on. Imagine a few iterations of this: the teachers are thinking "yup, the students all get it, we can even make the problems harder to get a better distribution of grades", and the students keep finding better ways to avoid having to spend 20 hours working through a complicated tangled mess of code. Are the students learning anything of value? Sure, they're learning how to play the system for their own benefit. They're learning that the appearance of being intelligent and successful is more important than the reality (see Sam Bankman-Fried, for example) when you're stuck inside a corrupt system. This of course explains the coding interview in the software industry: everyone knows this is how the current educational system works, so looking at grades, recommendations, accolades etc. is a waste of time. Instead they say, "Here's a problem, solve it in real-time using just the skills you've actually learned and the knowledge you've actually memorized." ChatGPT might make it a little worse than it already is, when it comes to ranking students from excellent to mediocre - but it's just a cheaper version of the private tutors that so many students utilize. It might even level the playing field somewhat, as tutors are expensive. Ultimately, the only real solution, assuming grading is important, is to proctor the students during real-time tests, with all their devices locked away. This makes work much harder for the teachers, particularly with large class sizes, as auto-grading based on online submission of work is so much easier.