cheating at Walls?

[Anonymous]

has anyone heard about rampant cheating at Walls? My Walls student says that there is "a lot" of cheating...but (predictably) clammed up when I started asking questions....

[Anonymous]

There’s a lot of cheating everywhere. ChatGPT is part of it. And there are a lot text groups where people send images of their answers. The lines between collaborating, checking your work, and copying can be blurry.

A friend whose nephew is not in DC recently told me about watching him photograph his assignment, having his phone convert the image to text, pasting the text into ChatGPT, then pasting the ChatGPT result into the assignment field and submitting. Complete essay, zero typing, zero thought.

Kids who do this stuff are ultimately cheating themselves, but many of them don’t understand that (and others don’t care).

[Anonymous]

Anonymous wrote:has anyone heard about rampant cheating at Walls? My Walls student says that there is "a lot" of cheating...but (predictably) clammed up when I started asking questions....

What grade are they?
My 9th grader has people who ask her for HW a lot, but doesnt see a lot of cheating going on

[Anonymous]

11th grade. Not Chat GPT or homework--tests. That's all I got before he clammed up

[Anonymous]

How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

[Anonymous]

My kid reports seeing it often, but we just tell them MYOB. From sharing HW to talking about tests and cheating online copy/paste other students' work. Some of those kids will seemingly do anything to get ahead. Sad.

[Anonymous]

Kids copy homework all the time at every high school. I haven’t heard about cheating on tests rampantly or anything but that doesn’t mean it doesn’t happen.

[Anonymous]

Anonymous wrote:How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

You clearly don’t know how to use ChatGPT 4. The Stanford head of admissions had it write admissions essays and he said they were in the top 1% of essays that his staff read (they didn’t know they were reading Chat essays).

Of course, you likely have no clue how to train the model…but just like to leave snide remarks.

[Anonymous]

Anonymous wrote:

Anonymous wrote:How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

You clearly don’t know how to use ChatGPT 4. The Stanford head of admissions had it write admissions essays and he said they were in the top 1% of essays that his staff read (they didn’t know they were reading Chat essays).

Of course, you likely have no clue how to train the model…but just like to leave snide remarks.

Oh no. Was that in a published article? I'd like to read more.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

You clearly don’t know how to use ChatGPT 4. The Stanford head of admissions had it write admissions essays and he said they were in the top 1% of essays that his staff read (they didn’t know they were reading Chat essays).

Of course, you likely have no clue how to train the model…but just like to leave snide remarks.

Oh no. Was that in a published article? I'd like to read more.

NP and college admissions essays have been written by people other than the applicant for years. Lots of parents write their kid’s essay, or have a private tutor do it or a private college counselor or whatever. People find ways to cheat with or without technology.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

You clearly don’t know how to use ChatGPT 4. The Stanford head of admissions had it write admissions essays and he said they were in the top 1% of essays that his staff read (they didn’t know they were reading Chat essays).

Of course, you likely have no clue how to train the model…but just like to leave snide remarks.

Oh no. Was that in a published article? I'd like to read more.

The problem is that it was in an interview with multiple AO's on a wide variety of topics, and ChatGPT was just one of maybe 5 different topics. I should have emailed the article to myself just to be able to refer to it because I was impressed by what the guy said.

Not the same, but here is from a Washington Post editorial by a University of Wisconsin professor:

Lawrence Shapiro is a professor of philosophy at the University of Wisconsin-Madison.

ChatGPT has many of my university colleagues shaking in their Birkenstocks. This artificial-intelligence tool excels at producing grammatical and even insightful essays — just what we’re hoping to see from our undergraduates. How good is it, really? A friend asked ChatGPT to write an essay about “multiple realization.” This is an important topic in the course I teach on the philosophy of mind, having to do with the possibility that minds might be constructed in ways other than our own brains. The essay ran shorter than the assigned word count, but I would have given it an A grade. Apparently ChatGPT is good enough to create an A-level paper on a topic that’s hardly mainstream.

[Anonymous]

Anonymous wrote:How do kids do well with a chat GPT essay?

Is that the standard of A essays at public schools?

This post is a great illustration of the Dunning Kruger effect.


The core function of LLMs like ChatGPT is to predict the next token in a sequence. In the case of ChatGPT, the first sequence it analyzes is your prompt, then it’s your prompt plus the previously generated token and so on.

So, the prompt determines the quality of the response. For example, if you prompt “write me an essay about the reasons for the US’s entry into the Vietnam war” you will get a very generic response. But you’ll probably get a much more interesting essay if you prompt “From the perspective of a highly intelligent and politically neutral history professor who is uniquely gifted in understanding the inner motivations of historical figures, write me an essay of approximately 2000 words comparing and contrasting the viewpoints of Lyndon Johnson and Ho Chi Minh on why the US entered the Vietnam war”.

Most people think that they can “detect” LLM generated text, but they think that because they are not good at writing prompts. With a little bit of practice at writing prompts, the output gets much better. Research shows that both humans and AI detection software are terrible at detecting AI output and that they have a huge false positive rate, particularly for writing by humans who are not native English speakers.