ChatGPT - dumbing down of education?

[Revoltingest]

I generally favoured these type of approaches when I was a teacher, but had to balance it against the need for my students to be able to handle other tests (standardised, entrance exams, etc) which would not be open book. So I did a little bit of the closed book thing, just to teach them how to do closed-book exams. Frustrating at times.

Yuck.
The real world is open book.

 

[Stevicus]

I wonder if anyone has tried the "who's on first" routine with ChatGPT.

 

[vulcanlogician]

The genie is out of the bottle, in many ways.
But there are also ways to minimise the harm of it, in an educational sense.

Which ways?

And I don't quite get why you provided the two essays. What point are you trying to make with them? (Sorry if I'm missing something obvious.)

Btw, if I were a university English teacher, I'd assign an A to a student who handed such an essay in to my English 101 class, but a C or D to a graduate student. But I suppose such a simple prompt would never be given to a graduate student in the first place.

The thing is, when I was in school, I NEEDED time to write a proper essay. My first drafts were trash (and often ended up in the trash can). But, given time, I was able to produce papers that scored 99-100%. What about THOSE students whose ability to craft good essays depends on skillful revision and multiple attempts? It seems like giving 45 minutes in-class to write an essay is going to make some students look like bogus writers when in fact they need time to sculpt their work.

 

[Stevicus]

I wonder if anyone has tried the "who's on first" routine with ChatGPT.

Did Abbott and Costello ever say who was on first base?

Yes, "Who's on First?" is a comedy sketch by Abbott and Costello that revolves around a conversation between two characters named Abbott and Costello. In the sketch, Abbott is trying to explain the positions of baseball players to Costello, but the conversation becomes increasingly confusing and hilarious as the two characters misunderstand each other.

The central gag in the sketch is the misunderstanding over the name of the baseball player on first base. When Costello asks Abbott "Who's on first?", Abbott responds that "Who" is the name of the player on first base, leading to further misunderstandings as the two characters continue their conversation. The sketch is a classic example of Abbott and Costello's comedic timing and wordplay, and it remains popular to this day.


If I had asked the question "Who's on first?" just by itself, would you have been able to get the reference to Abbott and Costello, or would that have been too vague as written?

As an AI language model, I would have recognized the phrase "Who's on first?" as a well-known comedic catchphrase, but I would not have been able to attribute it specifically to Abbott and Costello without additional context or information.

The phrase "Who's on first?" has become a popular cultural reference and is often used in various contexts to express confusion or a lack of understanding, so it's possible that someone might use the phrase without intending to reference the Abbott and Costello sketch. However, if someone were to specifically mention Abbott and Costello in the context of the phrase, it would make the reference clearer.

 

[Mock Turtle]

When I was in engineering at UofM, every test was
open book. All ya had to do was understand the
material...no memorization.
In some courses, we were graded on homework
that had results posted before it was due. This was
easier for me cuz I hated memorization.

I hear that it's no longer my utopia.

I'm not sure how useful it was to have memory as a requirement - fortunately it wasn't a problem for me - and I'm sure it did hinder many. Thankfully times change.

 

[Koldo]

Yeah, sorry. I was typing away on a couple of different things at the same time. You'd need to read my conversation here with sunrise to work out what the context was for my conversation with you. My bad on that.

Basically, I was making a point earlier that writing essays at school (as an example) isn't really about the topic of the essay. At least, not mostly, and not usually. It's mostly about learning how to structure a convincing argument, and evidence your position.

I find it interesting that you bring this up and then later on posts an example essay from ChatGPT that is both lacking in properly presenting an argument and in evidencing a position. I would expect students to create their essays with the same problems though, because what teachers generally expect, in my experience, is showing that you have researched about the topic, and that's all.

 

[Revoltingest]

I'm not sure how useful it was to have memory as a requirement - fortunately it wasn't a problem for me - and I'm sure it did hinder many. Thankfully times change.

One only need remember enuf details to know
how to look up what one doesn't know.
Concepts are the most useful thing to know.

 

[rosends]

One only need remember enuf details to know
how to look up what one doesn't know.
Concepts are the most useful thing to know.

While concepts are certainly important, one's brain is a muscle which needs a workout on many levels, including memorization

Dan Rosen's other Blog: LOTS of praise

 

[Revoltingest]

While concepts are certainly important, one's brain is a muscle which needs a workout on many levels, including memorization

Dan Rosen's other Blog: LOTS of praise

Me noggin got enuf of a workout just
learning the concepts & applying them.
Nowadays, I don't memorize things, but
I do enhance memorization by entering
info into me computer. And it remembers
for me when gray matter glitches.

 

[idea]

It is a fine balance, I remember the outrage when calculators were first allowed in exams. Shock - Horror.
There is plenty of software out there that can spot 'exact copy' and plagiarism; examiners learn to modify questions to explore the children's understanding and application of knowledge.

Yes, the church of TI and the 80's calculator wars. Poor Casio, schools are still forcing kids to buy expensive, copy-cat, dishonest, outdated 20yo+ $100+TI's from that manipulative company.

Say no to TI kids they are corrupt as hell.

Casio made the original, and is better for cheaper

 

[idea]

BTW, the short run for chat writing essays for students is almost over

Uh oh! Turnitin is officially going to detect if you've used an AI chatbot in your essay

 

[Altfish]

Yes, the church of TI and the 80's calculator wars. Poor Casio, schools are still forcing kids to buy expensive, copy-cat, dishonest, outdated 20yo+ $100+TI's from that manipulative company.

Say no to TI kids they are corrupt as hell.

Casio made the original, and is better for cheaper

Casio are accepted in the UK

 

[lewisnotmiller]

I find it interesting that you bring this up and then later on posts an example essay from ChatGPT that is both lacking in properly presenting an argument and in evidencing a position. I would expect students to create their essays with the same problems though, because what teachers generally expect, in my experience, is showing that you have researched about the topic, and that's all.

I'm glad you found my post 'interesting', but the context suggests you meant 'inconsistent' or something. Which is also fine, Ive got no dramas with disagreement, just trying to work out how to respond.

1. How the essays I posted as examples would be received depends on lots of things, including not only context and teacher expectations, but the level of the class, and the way the question is framed to chatGPT. I limited it to 800 words, but could have had a 3000 word version in basically the same time. I could add specific considerations (like, consider this in light of the Jan 6 insurrection) just as quickly, and punch out multiple versions.

2. I don't see ChatGPT as a tool to quickly produce collegiate level persuasive essays. Yet.
But it seems like this will change.

3. Essays might not even be the most immediate example. I tried getting it to write some code for me to do a specific thing in a specific language and it produced a decent codeset, also pointing out some basic error handling it lacked. I asked it to redo the code , factoring in the error handling it had noted as missing, and it did so.

4. I've taught at primary, secondary and tertiary levels, as well as trained adults in a professional setting. Majority of my professional experience is with 10-13 year olds. I agree this doesn't have much application as a cheat tool at certain levels of education. I think it does in some contexts, and suspect this will grow as the tool develops.

Hope that helps explain what I'm thinking, happy to further discuss, and (of course) feel free to disagree with some/all.

 

[lewisnotmiller]

BTW, the short run for chat writing essays for students is almost over

Uh oh! Turnitin is officially going to detect if you've used an AI chatbot in your essay

Sure. Unless the AI gets better.
Which looks pretty likely.

 

[lewisnotmiller]

Which ways?

Ha! I'm not quite suggesting I have the answers. More that actively considering it in designing coursework, and in particular homework, allows mitigation strategies to be put into place. Without too much of thought, this might include;
1) Use of GPTZero or similar tools to try and detect AI developed outputs.
2) Requirement of drafting works to be shown where development of a long text (eg. narrative writing) is being done.\
3) Requirement to present views on essays previously written in an oral setting without having the essay in front of you, but having access to your own handwritten cheat sheets or notes.
4) Increased use of in class assessment, as opposed to homework.

And I don't quite get why you provided the two essays. What point are you trying to make with them? (Sorry if I'm missing something obvious.)

Just playing, really. A more fun example I just did with my daughter in trying to show her how it works was to say to ChatGPT 'Give me a good recipe for pancakes which includes choc chips.' It gave me one, including ingredients and method. I then said 'I forgot to tell you I'm allergic to eggs'. It apologized for not checking dietary requirements, and then gave me a new recipe, which excluded eggs. It's entirely possible to throw together a quick result, then add more details and get a more refined result.

Btw, if I were a university English teacher, I'd assign an A to a student who handed such an essay in to my English 101 class, but a C or D to a graduate student. But I suppose such a simple prompt would never be given to a graduate student in the first place.

Yup, that's fair, and I think you're being generous. But my daughter (Year 7 student) was pretty impressed by the 750 word essay on the perils of social media I just generated in 30 seconds, and was pretty sure she'd get an A. Particularly when I regenerated it, and asked it to use Pew Research plus an Australian example and it did so.

The thing is, when I was in school, I NEEDED time to write a proper essay. My first drafts were trash (and often ended up in the trash can). But, given time, I was able to produce papers that scored 99-100%. What about THOSE students whose ability to craft good essays depends on skillful revision and multiple attempts? It seems like giving 45 minutes in-class to write an essay is going to make some students look like bogus writers when in fact they need time to sculpt their work.

I think this is a good point. I would say simply that ChatGPT should be considered in working out how to assign assessments. It might be as simple as saying 'Hey so don't throw out your drafts...submit them along with your finished product.'

Then you have more of a drama with the kid who punches things out on the first try, but THAT kid I should be able to see better evidenced in class.

I'm not suggesting I have a great answer, but the first step is to consider things actively, instead of ignoring them. (Not suggesting you're ignoring them...I know some schools here are at this point.)

 

[Koldo]

I'm glad you found my post 'interesting', but the context suggests you meant 'inconsistent' or something. Which is also fine, Ive got no dramas with disagreement, just trying to work out how to respond.

If anything, what draws me to this forum is the possibility of disagreeing with people without having to engage in drama.

I didn't mean to call you inconsistent, and I would have no problem saying so if that was the case. I found the situation interesting because the text you got from ChatGPT would have been given a low score if you assigned a significant value to creating arguments and evidencing them. I can elaborate if you want.

1. How the essays I posted as examples would be received depends on lots of things, including not only context and teacher expectations, but the level of the class, and the way the question is framed to chatGPT. I limited it to 800 words, but could have had a 3000 word version in basically the same time. I could add specific considerations (like, consider this in light of the Jan 6 insurrection) just as quickly, and punch out multiple versions.

2. I don't see ChatGPT as a tool to quickly produce collegiate level persuasive essays. Yet.
But it seems like this will change.

One thing I have tested is asking for sources when given an answer on ChatGPT. Which are mandatory at college level works. But... The one time I was given the source (a website) I couldn't actually find the original statement.

3. Essays might not even be the most immediate example. I tried getting it to write some code for me to do a specific thing in a specific language and it produced a decent codeset, also pointing out some basic error handling it lacked. I asked it to redo the code , factoring in the error handling it had noted as missing, and it did so.

4. I've taught at primary, secondary and tertiary levels, as well as trained adults in a professional setting. Majority of my professional experience is with 10-13 year olds. I agree this doesn't have much application as a cheat tool at certain levels of education. I think it does in some contexts, and suspect this will grow as the tool develops.

Hope that helps explain what I'm thinking, happy to further discuss, and (of course) feel free to disagree with some/all.

I most certainly agree with the underlying premise that this will have an impact on education.

 

[vulcanlogician]

1) Use of GPTZero or similar tools to try and detect AI developed outputs.

If such an AI/neural net could identify even 50% of AI generated texts it analyzes, it would enough to curtail the use of AI in academic writings. I hope this solution turns out to work.

I mean, look at it this way, PER ASSIGNMENT there is a 50% chance that the student would get away with it. But you can't graduate college only handing in one paper. Sooner or later, the student would get pegged for academic dishonesty. (And, in my view, that's exactly what using an AI to write with is.)

That's an interesting side question: is using an AI to write papers academic dishonesty? Of course you could have certain assignments in which the student is allowed to use AI, and in those cases, it isn't. But in (especially in the humanities) students are expected to do their own research, construct their own arguments, and organize their own writing. If an AI is doing this for you, (and you are supposed to be the one doing these things) I think it certainly does qualify as academic dishonesty?

What's your opinion on the issue of AI and academic dishonesty, lewis? I'm also curious what ChatGPT would say if asked.

 

[lewisnotmiller]

Picture if someone invented a set of bionic leg braces. You place these on your legs. They are slick, shiny and inexpensive, allowing anyone to run with world class sprinter speed.

This invention would be impressive to your peers, while owning a pair would good for your vanity. But does it strengthen your natural legs or does it just act like a crutch? If we took off the race braces, after a few months of running, to keep up with the Jones, would your natural legs be in better shape or worse; free running, than when you first put on the braces? I would guess worse, since your leg muscles would atrophy, like in a cast, due to lack of deliberate use.

I can give you a practical example from personal experience. Due to a considerable number of severe ankle injuries (thanks basketball) I took to wearing ankle braces on both legs.

Did this for years. And it helped in terms of short term injury prevention each game. What it didn't help with was long term strengthening of my ankles...quite the opposite.

Not wearing the braces was better for strengthening...but increased the chance of reinjury. Just a matter of competing priorities really.