AI chatbot doing kids' homework for them - undetectable

@Itsbiasedhere I would like to go back to when nobody gave a shit about pupils with SEN /disabilities.
There you go I rewrote your post for you no AI needed.

To be honest, I think it's most likely to be of value to lazy, but bright, students.

Too lazy (or disorganised, demotivated, busy etc) to do all the detailed research/reading, but they probably already have an understanding of what's required
BUT
bright enough to input detailed questions and evaluate the answers
THEN
augment/finesse with some extra detail.

So, from what I've tried, it can't write a BTEC assignment from scratch, just given the brief. If it's just given subheadings/prompts, it produces reasonable, if vague work. If you try to get it to draw conclusions from your data, you have to be quite specific in terms of what you ask it to get anything meaningful. I think my weaker students would struggle to frame their queries in a way it understands.

I do think it could help students achieve a pass, but I don't think it could produce a whole assignment for them. It's definitely something to keep an eye out for and be aware of, though!

Pretty much all level 3 BTECs I think- in part because it would be very difficult to allocate enough class time to do the assignments required. But for applied science there is generally a lot of calculation using your own data, talking about your specific experiment etc. I think there are assignments where weaker students could use it to drag themselves up to a pass, but for the majority of assignments, they'd have to understand the principles of what they are doing in order to do data analysis etc.

What it can do is edit their work for clarity- I tried putting in some text some of my weaker students have produced, and it can give the same content but expressed much more clearly- which would help some students.

I’ve been using AI for “fun and research” for a couple of years and haven’t been overly impressed.l with either its output or me not being paid for doing the ML! 😂

I’ve not read R&J (or seen the films) - but from what I could see, the AI chucked out basically 3 points and phrased them slightly differently. I think if I googled “give me a 2 minute synopsis of R&J”, I’d get a far more comprehensive answer!

The devil’s in the detail.

If I was a student using university wifi or browsing software I would be avoiding even visiting the ai chatbot website, never mind plugging in my essay questions in the name of "curiosity", so as not to give any reasonable grounds for suspicion.

Interestingly, a lot of university 'academic integrity' clauses don't yet cover the use of AI essay writing software, so it would be interesting to see a case being brought against a student for this.
Typically a student agrees that they will submit their 'own work' and not get 'anybody else' to write their work for them, or to copy 'another person's work'.
However, arguably, AI writers are none of these. A student can argue their 'own work' (submitting questions) has led to the output, it is not plagiarised, nor has 'anybody' else produced it.

If I use Word to spellcheck my essay, is that still my 'own work'?
What about if I use Grammarly or Quillbot to reword my text to make it more grammatically correct, or formal? Is that still my 'own work'?
Both of these are legit and were encouraged by my uni tutors.

It's a grey area and it will be interesting to see how it develops.

looking closely as the answers it produce, all are very lengthy with a lot of flaffing, not much substance. One possible application is politicians' answers to interview questions.

I haven't yet found a chatbot that is actually useful, so any improvement in them would be good.

they are offering beta version for free as baits just like any other platform. People are all amazed how incredible it is. All companies are now lining up to get the commercial version to improve their chatbots.

See here for the response from leading plagiarism detector Turnitin: www.turnitin.com/blog/ai-writing-the-challenge-and-opportunity-in-front-of-education-now

In summary, they can't detect it yet, but anticipate being able to do so soon (well of course, they must say that).

If I was a student using university wifi or browsing software I would be avoiding even visiting the ai chatbot website, never mind plugging in my essay questions in the name of "curiosity", so as not to give any reasonable grounds for suspicion.

I once had an essay talking about spy equipment and where you could buy hidden cameras. This was in health and social care, they had been asked about 'health surveillance' ie film badges id you work with X-rays
This is brilliant 😅

Oh definitely - ds used a laptop with no internet enabled (dyspraxia).

It's about being able to show organic progress - here is Exhibit A, a supervised draft we can confidently authenticate - here is Exhibit B, a more polished final draft the student has worked on independently (& it's up to them to decide whether to give it a cursory tweak or slave over getting it just-so)...here is Exhibit C, the formative feedback from the teacher/supervisor which has enabled the student to move from A to B.

Not 'no essay' --> 'sudden appearance of final product which has sprung up like a mushroom overnight'!

The examples posted upthread are pretty rubbish, anyway, but presumably the tech will improve.

I'm just saying that commissioned essays have been absolutely standard for well heeled students since forever. This AI, if anything, is levelling the playing field so that everyone can cheat for free...

Both the IGCSE course, IB course and IGCSE courses that we teach have coursework in them. We have oversight in terms of planning and drafting over a long period of time though and we know our students well so it’s pretty easy to have oversight.

Ha - me too, actually. Well, no ADHD diagnosis although I have my suspicions (dd2 has ADHD) but yes, I too often bang out entire schemes of work overnight in a hyperfocused deadline panic!

So no, not the only criterion. But if you're supervising an extended essay for IB - which requires considerable pre-reading & preliminary research & planning - & you know perfectly well the kid hasn't even started this because you've had a progress meeting where the kid just grunts 'dunno' to everything - then a rather polished preliminary draft on a similar, but not the same, research questions drops into your inbox at 4am...it's viva time.

We can tell.

When a student who struggles to write coherent English suddenly hands in a perfect essay you start to get suspicious.

The 'tippex test' is also useful. Copy the essay, tipex out every 5th word and ask the student to fill in the blanks.

Some students don't even check the font so the essay has 4 different ones.

I once had an essay talking about spy equipment and where you could buy hidden cameras. This was in health and social care, they had been asked about 'health surveillance' ie film badges id you work with X-rays.

Hopefully this means the end of coursework for a levels and GCSEs I'm sure teachers don't want to mark it and I'm sure it was a high cheat risk anyway. I would like to go back to ye olde traditional exams only like in the 60s and 70s.

I think our educational obsession with exams as the only means of testing students’ intelligence and ‘worth’ needs to change. The learn and regurgitate model really only assesses one set of abilities and prevents a significant proportion of students from demonstrating their capabilities. Moreover, it’s a skill which is rarely used in life after school - the ability to access information and evaluate it, build on it etc is far more important these days.
I understand teachers are concerned but I do think the solution will be to engage with students abut AI and evaluate its strengths and weaknesses.

Students can't do this in an exam though so if they don't do any work and just get the Ai to produce their essays they aren't learning. The only thing that should happen is that no coursework goes towards the final mark.

The work was rejected largely because it didn’t meet the brief. It was only a handful of weaker students anyway, but their responses were both different enough to be unique, but stylistically similar and definitely not their own work (I told them I knew this, and they didn’t deny it). They resubmitted with some intensive supervision and achieved passing marks (just).

On what basis was the work rejected? What reason was the student given?

TurnitIn won’t pick this up because it’s essentially generating ‘new work’ not cutting and pasting from elsewhere. What I’m not clear though is if it always generates the same answer for the same question? So, if a class of 20 all asked the same question, would the responses be the same?

I’ve been teaching international foundation and first year HE students over the past year and am pretty sure I’ve seen this in action, was pretty obvious (and the work was rejected) though I wasn’t sure at the time what it actually was.
Interestingly Turnitin wasn’t picking it up - it just read very strangely and at odds with what I knew of students’ abilities from class. References were “off”.

Agree with @Piggywaspushed - in my subject the final grade all goes on final exams with no coursework and so we spend many an hour training all students to perform to the very best of their abilities in timed, closed book exam conditions, taking into account any SEND requirements and working with those rather than against them by developing exam strategies for each student.

I wouldn't be teaching professionally or prepping my students for exams properly if I didn't do that.

I have ASD and ADHD myself so I'm fully aware of the organisational and practical accommodations that often need to be in place in the classroom and the exam hall.

Students with the SN you describe would have access to word processing in in class assessments. I'd imagine most schools will block access to this site.

As far as handwriting hem assessments, I always have this as a choice and the take up is usually about 70:30 in favour of typing longer pieces. After playing about , I am not worried about this ChatGPT for my subject at all. It's failed every task I set it!

To all the teachers saying they will switch to handwritten, in-class assessments to combat AI like this, PLEASE consider the impact of this for students with SEN.
DS (dyslexia, ADD, slow handwriting (poss dyspraxia?)) would fail massively under this regime at school, and yet is now working at 2:1/1st level at uni.

Also, I was absolutely the sort of student who would produce work overnight at the last minute once the deadline adrenaline kicked in!

I have adhd - unfort developing a 2000 word essay overnight is a common adhd trait as we leave things to last minute and the pressure allows us to trigger our ability to hyperfocus. So please don’t consider that as the only criteria to suspect someone has cheated.