If I was a curator on HIVE, I'd be taking notice of Chat GTP and the authenticity of writing here now that this AI language tool is here. In fact, there isn't a profession that shouldn't be taking notice of it, and as the school term begins, many teachers are talking about it's capabilities. Much to my shock, the principal at Jamie's school hadn't even heard of it, but she didn't realise that vaping was a problem either, which says a lot about her cabilities.
A quick play this morning and we have got it to write, in a less than a minute over coffee:
- a lesson plan on Newton's Universal Law of Gravitation
- a 200 word analysis of symbols in To Kill a Mockingbird
Problems?
Yes, but nothing that couldn't be overcome by redefining the parameters so the AI knows exactly what you want. It does take some skill and practice, but by the end of this blog post I was getting some quite good responses, from essays to lesson plans, creative writing, poetry and more. I even got it to write a reasonable rubric.
Thinking back to being in the English classroom, there's a definitely things I'm not overly concerned about, and a way to use Chat GTP as opportunity. There's no way English teachers are going to ignore it, and I don't doubt there will be much discussion, analysis of criticism and media analysis on this new and world changing technology.
To begin with, students need a degree of literacy to use it. I used to watch students struggle with searching for things on Google most of the time, and that tech has been around forever. Using specific search terms is something we have to teach, and often, to save time, I find information for students before the lesson is over and no work has been done at all. To use Chat GTP adequately, high school students would have to be taught how to use it. The top students might be that clever, but the top students don't need to cheat.
For example, 'write a 200 word analysis on 'x' text' is fine, but unless they add modifiers such as 'add quotes and examples from the text', their work would not get higher than a C, perhaps a B, if it was particularly well executed. A student who usually writes at a lower standard would be immediately flagged for writing something in the way CHAT GTP is capable of, and monitored closely for cheating.
Students would also have to memorise an essay before a handwritten, timed exam, and know the answer to the question first. They would have to ensure they adhered to the criteria they would be marked upon.
For example, when I asked it to 'write an introduction to an essay on 'Lord of the Flies', it arrived at this:
In the novel "Lord of the Flies," William Golding explores the inherent evil of human nature through the story of a group of boys stranded on an uninhabited island. As the boys struggle to survive and govern themselves, they begin to lose their sense of morality and civilization, ultimately descending into savagery. Through this gripping tale, Golding raises important questions about the nature of society and the potential dangers of mob mentality. The novel, first published in 1954, remains a classic of modern literature and continues to be widely read and studied today.
Not bad. however, 'gripping tale' and the publication date, as well as the fact it's widely read and studied, are irrelevant to the criteria and superfluous. It is also general, sweeping over the main themes. I would comment that the student needs to closely answer the question - however, as English essays in senior classrooms here are given on the day and not beforehand, the student would be hard pressed to know what the question was about.
If the question was 'How does Golding create a sense of foreboding in the novel', this introduction would not answer the question at all. Better, though, if you know the question:
In the novel "Lord of the Flies," William Golding masterfully creates a sense of foreboding and tension that keeps the reader on edge throughout the story. Through the use of imagery, symbolism, and characterization, Golding paints a picture of an uninhabited island that is both beautiful and dangerous, and a group of boys who are struggling to survive and maintain their sense of morality. The novel raises important questions about the nature of society and the potential dangers of mob mentality. In this essay, we will explore the various literary techniques that Golding uses to create a sense of foreboding in "Lord of the Flies" and how they contribute to the overall impact of the story.
Still, 'keeping the reader on edge' is informal vocabulary, the introduction is far too long (they still have a whole essay to write in an hour) and it is a little repetitive, with superfluous lines. No 'short,precice, succinct' as I would advise students sitting VCE assessment here in my home state in Australia.
Can CHAT GTP used positively in the English classroom?
Without a doubt, new and concerning technologies can be used in positive ways in the classroom.
It's quite fun to get the Chat GTP to 'write a poem based on the theme of' or a poem based around an event, a character or chapter. A ten line poem on Macbeth?
The ambition, it burns like a fire
It drives Macbeth to his desire
He longs for power, and a crown
And will do whatever it takes to get them now
But the cost is high, a steep price
As he commits murder, twice, thrice
His ambition, it clouds his mind
And leaves him alone, in a state of grind
In the end, he loses it all
And his ambition, leads to his fall.
Of course, another opportunity to question and assess. What is a 'state of grind' and does this apply to the play? How many people does he kill exactly? Does the AI do a good job at summarising the play? What is left out?
Another lesson could be to expand on something the CHAT GTP has begun. For example, 'write about a haunted house in 50 words'.
The haunted house on Elm Street was infamous for its eerie noises and strange shadows. Locals whispered about the malevolent spirits that lurked within. Brave souls ventured inside but were driven out by ghastly sights and sounds. Empty now, the house still holds a malevolent presence, waiting for its next victim.
Students could:
- assess the response according to the set criteria
- list verbs, adverbs, nouns
- define unknown or difficult words eg infamous, malevolent
- pair up and get students to suggest a way in which this passage could be expanded upon, for example: desscribe a local's experience in the house, use three sentences to describe the noises in the house.
- after doing so, they could ask it to complete the activity itself and then compare to their own response. In what ways are they similiar/different? How might the Chat GTP be improved upon? In the below example, I would think that students would immediately identify that the repetition was unnecessary.
- Students could then rewrite the passage with their own improvements.
The ghastly sights and sounds inside the haunted houseon Elm Street were said to include ghostly apparitions that appeared out of nowhere, blood-curdling screams that echoed through the halls, and strange, unearthly noises that seemed to come from within the walls. Some reported seeing furniture move on its own, and others claimed to have been touched by unseen hands. The air was said to be thick with the stench of rotting flesh, and the shadows seemed to shift and writhe in an unnatural way. Some people even claimed to have seen ghostly figures of previous occupants who had met a tragic end within the house.The ghastly sights and sounds were so terrifying that many who ventured inside were too scared to ever return.
Arguably, students need a degree of literacy to do any of the above tasks. A higher level literature classroom would have no problem and perhaps find the task exciting, yet a lower level class who struggled with literacy might not be able to understand the task or struggle to read the content and identify it's failings according to set criteria.
An opportunity for critical thinking
One of the main opportunities that has been flagged for negotiating this AI in the education is around critical thinking. In my experience, education has removed critical thinking in favour of producing generic responses that get good marks. Students are often worried about making their own assessment or judgement of literature in favour of the 'right' answer in study guides. As English teachers and assessors, we are looking for work that questions, interprets, evaluates. How might education take this opportunity to design assessments that are outside the realm of ChatGTP and still show understanding and assessment of texts?
I have long argued for a portfolio assessment in the English classroom, extending over the term where the students write various responses to the text based on their own interests, and with close conversation with the teacher about their thought processes. If essays can be written by AI, how might assessment change so that it assesses the human response?
An opportunity to save time
Teachers are always time poor and overworked. Whilst experienced teachers could create a lesson plan in their heads on the way to the classroom, sometimes creativity can be lacking or newer teachers are stuck for activities.
I wrote 'write a lesson plan on the theme of ambition in macbeth' and it was pretty good. There were bits I would remove that I don't think would work very well or be too time consuming but it was a good start. Clearly, a lesson plan is only as good as the teacher who critiques, adapts and uses it, so I can't see the Chat GTP replacing teachers just yet in all areas!
I can't even imagine how time saving this would be - to me it's mind blowing. Here, I got it to write a table rubric assessing the student's ability to understand the theme of ambition in Macbeth. Any good English team would sit and edit this to make sure it worked, but have you any idea how long even STARTING a rubric can take?
This is an absolute game changer.
It's here to stay
The Pandoras box has well and truly been opened. It's been coming for a long time, and any teacher who hasn't been thinking about how to approach it will find themselves threatened and worried.
Education has a great opportunity to embrace this new technology and well as talk to students about what it actually means for their ability to navigate the adult world in terms of their careers and even their intelligence.
Interesting times.
Oh, and it even wrote me a school report for someone who is lazy and needs to do homework:
Student John Doe is struggling in English class due to lack of motivation and poor study habits. The main issue identified is his lack of effort in completing homework assignments and participating in class discussions. Despite multiple reminders and opportunities for extra help, John has consistently failed to hand in homework assignments on time and has not been actively participating in class. This has resulted in a lack of understanding of the material and poor performance on quizzes and exams. To improve, John needs to take more responsibility for his own learning and start taking active steps to improve his performance, such as completing homework assignments on time, actively participating in class, and seeking extra help when needed.
Interesting times.
If you're a teacher, are you worried about CHATGTP in the classroom? Have you used it yet or discussed it with your students? If you're not an educator, have you used it yet? To what degrees of success? How do you feel about this AI?
Creepy images of the English classroom co-created by me with Midjourney AI. For the record, my classroom never looked like this.
With Love,
Are you on HIVE yet? Earn for writing! Referral link for FREE account here
I'm pretty sure we have a lot of AI generated shit on Hive already, I noticed this at the end of last year, and am seeing more and more including from at least one old timer. They're pretty easy to spot especially for a native English speaker plus if you look at the user profile and their comments, their post does not fit the profile. I'm gearing towards supported trusted people only and that means my circle of friends. Tough, but I really can't trust a lot of people now.
btw, have you heard of this? It's pretty accurate so far, confirmed most of the shit content I've suspected. As an English teacher, you can get yourself registered
https://gptzero.me/
Absolutely. I have seen it a lot on here, even from people who should know better. The temptation to use it (and abuse it) is too much for some people. On the plus side, chat GTP will become pay as soon as it is out of beta. On the negative side, I'm sure many other AI bots that are free will appear in the future, so the problem will continue. And that will get better and better, so it will become more difficult to recognize them, especially when we get more personalized AI bots that can be trained with one's own writing to better mimic it. Can you imagine? For example: I have 5 1/2 years of daily writing on Hive, and another 20 years of writing on my blog. A good AI bot could be trained on all that to completely mimic my voice. It is going to become really hard to tell bot from human in the future. I don't know if services like the one you link too will always be able to help as AI improves more and more.
I believe AI generated content has certain traits and styles which can be detected either by a human with strong language capabilities and an AI detector. A human will always have an upper hand because we can take into account of other factors and make a rational decision to err on the side of caution and dismiss a piece of work or accept as genuine depending on the entire account or student, is a teacher's case. It will be a race between AI bots between both sides on who can outwith the opponent, let's hope it will be the anti AI detector who wins!!!
btw, it's great news that Chat GPT will become pay only after beta, I wasn't aware of that.
I hope you're right, but I don't know. It will be interesting to watch the race between AI bot and AI detector, at any rate.
Hive is small potatoes (right now). What will really be interesting is when high level magazines like the New Yorker run into this problem. I wonder if they have already had any AI written submissions yet...
I wish I had the skills to train the AI on my voice! That would be fascinating. I am with you - sounds like you have been investigating as much as I have and know what it is capable of. I just got it to write pretty good, funny film reviews for Twitter. No way could you tell it was Chat GTP. I wouldn't use the, but it's a point.
I'm sure we will be able to personalize our own AI bot in the near future.
Related: Several years ago I read a silicon valley futurist claim that in the future we will never really die. That is to say, we will, but our kids/grandkids will then feed all of our writings (journals and whatnot) and all our photos too (as this will be holographic) into an AI bot and it will then "become" us, adopting and mimicking our personality. So "we" will always be there for our descendants. I thought that was pretty far out there when I read it, but seeing how AI is now developing, I can believe it will happen.
It's crazy isn't it. We are old enough and well read enough to know the stories of the future and how they are coming true. It's like speculative sci fi cant exist as it did as the future unfolds exponentially and is here, now. I was reading about how to get the AI to scan and learn from our personal writing but it's beyond me for now. I expect it'll be a paid service before long. Makes me wish I knew a lot more about coding etc. Sure, I can ask an AI to write code for me but I think it's a little more than that. 🤣
I tried to get it to write another paragraph of my story I wrote this week, and fed it what I'd written. It was kinda repetitive and didn't have my style at all. I have had more success with a story 'in the style of Asimov' which was quite interesting.
I doubt anyone will have writers block ever again 😂
Fascinating, fascinating stuff.
Thanks for the tool.
An old timer??? Fark. That's disappointing.
If they are non English speakers I'd definitely look harder. There's also very repetitive patterns too. Bit goodness for those who can promote engineer, I reckon it'd be harder to spot. Insane. Another problem for Hive.
There has been an attempt to create AI detector but they are not too accurate yet.I tested it with article on hive from @cmplxty and it passes on as written by AI 😂 though I am 100% he's not.
The problem is, AI and combined brain could be pretty powerful combination. A lot of people are using it nowadays but just to generate ideas or perfecting them. However, there are outliers here and there that fully write the article from AI because they're good with giving prompts. Sometimes it feared me the most when one is accused of using AI despite that it's just the way they talk. Some people like to use complicated words, verbs, and perhaps run-on sentences that could seem AI like but it's actually human. However, the are a few markers here and there that could make a writing passed on as AI and you also have to check the history of the person/the students.
The way I see it is that, writing but also defending their ideas vocally would probably shows their competence and see if their writing isn't AI or AI written. The thing is, AI can also summarize and refine sentences that also partially works like Grammarly and that is something acceptable I suppose as people also suggest to use grammarly to refine their sentences and make it more grammatically correct.
See my reply to livinguktaiwan below. I think as AI improves and as it starts to becomes more personalized (ie we train it on our own writing so it adopts our "voice") it will become nearly impossible to detect.
Ah yeah, we seem to forget that part where we can train AI to adapt our voice. Anyhow, ChatGPT is only fraction of All-things-AI as there are many tools out there like copyai, anyai that helps with content creation side of thing and people use it for commercial purposes too.
Yeah, this will only accelerate. It's always hard to realize you are within a massive social change in the moment, but.. we are in one now. By this time next year AI is going to be everywhere.
Ha that's interesting. In the twelve hours or so since writing the article I've spent more time on it and watching you tube vids etc and my view has developed further. Like the image AI, it can be co creation - I played with writing an article where I fed it my concepts, adjusted the prompts to fine tune, added a few sentences and wofds, deleted some, and voila, a half decent piece i don't ghink people would know was AI. The skill is in knowing what's a good piece or not and adjusting the AI for best result. Now that's not quite stealing, right? But I guess the curator discussion might be talking into account whether people are happy to reward something AI written when they themselves are spending time being entirely AI free. Interesting!
I really don't like the way that AI is going. Sure people might say that it has a place...but things like this are bound to be abused and with the education systems around the world already being dumbed down, I think it's a slippery slope. Here on Hive there are already such huge issues with plagiarism, that I can see these AI mechanics becoming an even greater problem as we go along. Maybe I'm old fashioned but I don't think it will be used for any good purpose to be honest.
Yes! I agree with the dumbing down. I worry we won't even produce kids literate enough to use it 🤣
I taught as computers and tech started to appear in the classroom and was a big advocate for them. By the end of my career, I was advocating they go back to pen and paper and was teaching critical thinking more than ever.
Note I'm not retirement age, I just quit due to stress etc.
Shaking my head, sadly.
What have we done?! @riverflows
I saw a story about this on the news earlier in the week and have to wonder what the impact will be on college admissions essays as well as everything that has to to with teaching English and writing. Pandora's box indeed.
It certainly is concerning. Government will be slow to act. But in crisis there's opportunity... The creatures ARE out of the box so I guess we have to assess where to next as the centre cannot hold!
I have downvoted a couple of posts trying to pass it off as their own work. I think it's quite ready to spot unless you are super good with what you ask and then customise what you get.
The thing I spot the easiest is the generalities it inserts.
Still, this is just the beginning. They are only going to get better.
Yes! You have to be a good prompt engineer which is a needed skill now! I'm getting better at it. AND you can tweak and add stuff. I could spot it I think, but I could also edit it to be unrecognisablY AI. But then, @meesterboom, we are intelligent and literate. Perhaps this is the message students need to get - you still need literacy to work the AI.
I THINK our writing is distinctive enough for all the alarm bells to go off if we tried to pass off work as AI 🤣 BUT it wont be long before it gets close. I'm still experimenting but have got to point that I realise tbere isn't much it can't do.
Even try to be funny.
Interesting read with good points.
All I can say is that I'm not a teacher in the educational system any longer. It feels like even more of a headache to grade and check students' tests and homework nowadays, although I feel this AI thing could be an advantage for prepping our own work, especially for people like me (you?) without much focus and too many thoughts.
xx
Totally. If I was still teaching I'd be using it for everything 🤣🤣 I just it to write thirty essay questions for a text and it did so in seconds. Imagine how long that would have taken irl.
I would have used it for my Uni thesis, would have saved me some years of suffering/ depression ;^)
Totally. I just used it to write a plot synopsis for a sci Fi. Kinda freaked me out.
What have we done?
Yesterday I was talking to a friend about my short experience using ChatGPT and my first impressions, and one of the things I told him was that it had limitations in informal language contexts where the ChatGPT language would raise at least suspicion.
In addition to having some criteria and some knowledge on the subject to correct ChatGPT, especially in redundancy and superficiality in certain topics.
But I think we're just at the beginning of this and it's probably going to get better in the short term, so it's amazing what's ahead.
So far I have found it quite efficient for structuring narratives, and this as a tool is certainly great!
Very interesting read. I haven't tested the chatGPT yet. I'm no teacher but don't you think it would be noticeable somehow? I mean, the 'style' must be somehow different from what the student would usually write? Or do you mean some students would use the chatGPT all the time? And as a teacher, you would never get to know the student so to say?
But I'm just trying to think back on my own experience of Swedish (I know it's not the same but that's the subject I'm comparing it to as Swedish is my native language) throughout my school years. I understand a lot has changed since then. And I don't mean just AI-wise.
(If it were the subject English here in Sweden I think the teacher would be suspicious because the student would then write 'perfect' English, and that's not expected).
Oh yes, if students were stupid enough to hand in something way above their own language level, we would spot straight away. I believe higher level students will be better at using it.
There's so much capability here. More so that the image generators.
Really interesting to read a thoughtful analysis of how ChatGPT could affect education. Maybe ChatGPT should be used to write the news. Maybe there would be some real analysis 🤦♂️ So much of the news on ChatGPT is just mindless drivel.
I think they are already using it to write news and blogs etc. Copy writing is dead. I couldn't trust anything on internet not to be AI these days.
I like your ideas for how it can be used in the classroom. This idea of embracing it and using it as a teaching tool is something I have also been telling people.
Most of my friends in education tell me that the essays it spits out are full of errors and can be easily spotted by a teacher who is paying attention. Or some variation of that argument. My argument to them is not only what you write (that with better commands, it can produce better text) but that this is just the beginning and AI will improve much more from here, so just saying "we can be smarter than the AI and recognize it" is very shortsighted and not very useful.
This will radically change education. It will be very interesting to see what it leads to in the years to come.
I hope it does radically change education. I feel that the pandemic and now this are forcing us to do just that, but some are digging their heels in. There's a lot of teachers out there that are too lazy, tired, stupid of time poor or all of above to manage this new world and are certainly afraid. It's a big shake up.
Prompt engineering is definitely the skill of this new age. The brighter kids will do it easily. Others will not have to wait long for the AI to get better.
So far the essays I have got it to write are pretty good. If you didn't know the student or they were good writers to start with, you might find it hard to pick. I found it better at passage or poetry analysis... Feed it a paragraph, say, and prompt 'analyse the theme of the horror of war" or 'analyse the use of light and dark" and it works very well. General prompts create general responses.
It's so fascinating I almost wish I was back in English staff meeting this week to talk about it with colleagues. Almost. My hubs teaches Physics so that's interesting too. I played with it for hours last night - I believe I could write a study guide for a test in a week with the help of AI. I MIGHT even try to do so.
Yeah I hope it changes education too. We need a new model. I mean the current one is more or less the Prussian model of two centuries prior with patches and tape to keep it going in the modern world. We really need a new model of education. If AI forces the change it will be doing us a huge favor.
I did the same with it. I chatted with it about Japanese history, haiku, poetry in general, theater, kabuki, literature, and philosophy. Needless to say I could catch many errors, but I was impressed nevertheless, especially in that it learned when I would correct it and it remembered the corrections, so it improved. It was most impressive with code. Well, I haven't coded since uni so I'm out of date, but I put whatever it gave me in xcode and compiled it and it ran! It also gave me a lot of instructions for accessing HiveSQL using python, but I didn't try to verify that. All in all, very impressive bot.
Wow, I wish I knew more about code to get it to write code, if you know what I mean..
Seems like you are a bit like me, going straight to literature to play. I tell you what it can't write good poetry for shit 😂 Perhaps haiku is a little different, or perhaps my prompting isn't very good. The fact it learns and remembers is so amazing.