NEA - ChatGTP

The teachers will have suggested he look at the work of Joe Bloggs and Anne Smith, as they are the leading historians in this area, and to make sure he explores factors x, y and z in his work.

If his essay sounds a bit weird, plus completely ignores those historians and looks at factors f, g and h instead, then teachers will smell a rat.

@MrsBennetsPoorNerves Apologies I was talking earlier to my family member of the academic type on the subject. I was told, in no uncertain fashion, that I'd got the wrong end of the stick. Sorry about that. @Sparklfairy knows more about it.

slinks off in ignorance

It isn't automating the process because it makes up citations etc. It's not a useful resource in this sort of scenario.

@Sparklfairy again very interesting. Thankyou very much for your post. Must watch this space given the ramifications to students and then the ensuing fallout.

Sorry, I wasn't involved in the testing, I just tested it myself! I do some freelance writing occasionally, and the industry is split right now between people panicking that Chat GPT et al are going to decimate the need for human writers, and companies worried that their writers are 'cheating' by using Chat GPT to churn out work faster and make more money.

I've heard of people getting their work thrown back saying it's AI written, so I ran my own (written before Chat GPT came out last year and I hadn't even used Grammarly etc) through it. There's a scale of 'very likely AI, likely AI, neutral, likely human written and very likely human written. Some pieces came back with very likely human written, some only 'likely' which can be enough for some companies to not pay or threaten to sack you.

I haven't had any problems with accusations but IMO it's impossible for them to prove whether it is or isn't AI that wrote it. Multiple software programs can come back with different results. The problem full-time freelancers have is they can lose work through no fault of their own.

The consequences could be bad for OP's son as this technology is so new; if the exam board blindly accepts the results of the software through ignorance I'm not sure he'd have much argument. Even Google has had to accept that they can't accurately detect AI articles for search engine results and have now changed their criteria for rankings basically allowing them 'if it's likely to benefit the reader'.

Yes, but so do many humans! 🙂

Whether it’s good at what it does is irrelevant. Whether it is the stuff of our future (I suspect it is and it will be the end of many jobs as we know them now) is irrelevant to this thread. The point is that it has no place in schools when kids are either doing homework or doing assessed coursework. It’s simply cheating. As such coursework is likely to become a thing of the past.

Good intro:

zapier.com/blog/how-does-chatgpt-work/

It spits out words, not knowledge.

That's also it's weakness.

It's only as good as the material it feeds on. The user has no control over it. Nor do I imagine that most users have the slightest clue how it actually works.

Google the search engine, historically has a blindingly simple algorithm. It lists websites in order of how many other websites link to them. It's elegant simplicity at its best.

I'm not so sure of this one. The user cannot audit or quality check either the material that it's feeding on or, the code behind it.

I'd stay well clear until it's more transparent

Very interesting @Sparklfairy Do you mind if I ask how you got involved in the testing?

But it’s the automation of the process that’s the issue

It's the automation that's the power of it - that's why it's called "artificial intelligence" - it's getting closer to what a human brain does, and in some cases will do a better job than some human brains. Schools need to acknowledge it in their policies and make it clear what it can and can't be used for, because it's not going away.

They can have software, but it's fairly useless. ChatGPT have even developed their own detection software and that's not 100% reliable either! I've tested it myself.

A big problem is it was always okay to use tools like Grammarly and Quillbot to help you reword and correct stuff, but these tend to have a certain AI style of writing/tone which the detector can be overzealous with. Even material I've written with no tools whatsoever except my own brain, detectors have come up with a score of less than 100% human. Not much less, but not 100% certain it's human.

But it’s the automation of the process that’s the issue. What is the point in a teacher giving a grade for an essay if a student has literally just typed the essay question into the programme and hit the generate button. They’ve learned nothing from the process and the result shows nothing about their understanding, their eloquence, their ability to pull out the important facts and discard unreliable information etc.

"Research" for essay writing has always been about reading other people's work, pulling out relevant points, re-writing them in your own words, and organising them in a structured way. ChatGPT is just automating the process. But it can't "sense-check" as well as a human can, and it can only ever be as good as the source information it uses. It will get better in future, and have lots of uses. Schools need to have clear policies on it - their students can't become the AI native experts of the future if they're not allowed to play with cutting-edge AI tech at school, so schools shouldn't block it, but obviously history students shouldn't be using it when writing their assessments.

Because we have to report malpractice:

inform the awarding body immediately of any alleged, suspected or actual incidents of malpractice or maladministration, involving a candidate or a member of staff, by completing the appropriate documentation

It’s an amazing tool and I use it in class as a teaching aid. My MFL students have tried submitting essays written by it but it is very easily proven by a phd style viva - if they’re unable to explain verbally ANY of what they’ve written. Of course MFL cheating is much easier to detect than history though.

What I don’t understand is how seeing proof of him using it is proof that he used it for that particular piece of coursework. How the school can report him without hard evidence.

I don’t think you can use the teenage brain argument that far. If your child is intelligent enough to write an a level history nea, he is intelligent enough to understand he’s being sanctioned for cheating.

I only made the teenage brain point to emphasise that they make silly decisions when teenagers.

Has he signed the candidate declaration?

If they were using school laptops to do NEA work in school then all the arguments about parental help, places to work at home and so on are pretty stupid. It seems like the school have given them the time and resources to do the NEA work in school so it’s irreversible whether someone else has their own desk etc.

ChatGPT exists and it can be a useful starting point when you’re stuck with a blank page but NOT when the task is 100% about your own writing, ability to draw on sources and come up with your own line of argument. Anybody whose ever met a teenager knows that as soon as school block one site they find a way round it, so that’s not much of an argument either.

Can't you just get him to log on in front of you and show you?

Clearly it's cheating either way though. I've just as a test put in my DC's history NEA title to see what it brings up. It gives a basic structure of a decent, well worded and grammatically accurate essay. It saves literally hours and hours of work. That isn't research, that's outright cheating. I've asked it various work related technical questions too. It brings up the answers. That isn't research however you try to spin it.

I really feel for you. And for him. Kids do stupid things. They don't think things through properly and their brains are not fully developed. The most important thing is that he learns from this and you as his parents help him to find a way through and get to where he wants to be rather than to have this become a defining moment in his life where everything seems hopeless and overwhelming. Thats a dangerous place for teenagers to be. He's lucky in that he only needs low grades to get onto his course (if I've understood that correctly)

Fair enough.

It's not even safe for that. It's still at the bit of fun state of devil

This is a learning experience for him- don't cheat.

I'd be encouraging him to face the consequences.

Have you discussed with him why what he did was wrong?

If he’s used it to research it’s not much different to using google and I wouldn’t consider malpractice. If he’s used it to actually write a response then it is. The school should investigate thoroughly before reporting.