| Pages in topic: < [1 2] | Off topic: ChatGPT and multilingualism Thread poster: Maria Teresa Borges de Almeida
|
|---|
Lingua 5B wrote:
We are talking about essays here, not translations.
I wasn't!
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
No!
| | | |
Lingua 5B wrote:
They can add a few mistakes but can't change the entire flow and subtext. Impossible. We are talking about essays here, not translations.
The only time I can imagine this being successful is in the case of new student, maybe transferred from another school, who submits AI-assisted work from the start, so you have no standard of comparison.
The same would true for a teacher who starts lessons with a new class for the first time. I think that recognising the "style" of AI-generated text is going to be an important skill, and it's one I've been trying to develop so I can apply it in my technical editing work.
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
Passage I
I settle on the couch ...
Passage II
Children and adults alike ...
Passage III
When we arrived there ...
OK, here goes. I think the first one may have had some AI input. That's because it's so bland and characterless, whereas the other two, especially Passage 2, have attitude and personality - something that is lacking in most AI-generated texts that I've seen.
| | | |
Philip Lees wrote:
I think that recognising the "style" of AI-generated text is going to be an important skill,
1. Surely the technology will improve faster than we can keep up?
2. If AI/MT is to be considered a suitable tool for translators, why not for students and everyone else? Isn’t this like barring them from using a computer?
| | | | Lingua 5B 
Bosnia and Herzegovina
Local time: 13:35
Member (2009)
English to Croatian
+ ...
Ice Scream wrote: Philip Lees wrote:
I think that recognising the "style" of AI-generated text is going to be an important skill, 1. Surely the technology will improve faster than we can keep up? 2. If AI/MT is to be considered a suitable tool for translators, why not for students and everyone else? Isn’t this like barring them from using a computer?
More like barring them from using their brain.
Philip Lees wrote: Lingua 5B wrote:
They can add a few mistakes but can't change the entire flow and subtext. Impossible. We are talking about essays here, not translations. The only time I can imagine this being successful is in the case of new student, maybe transferred from another school, who submits AI-assisted work from the start, so you have no standard of comparison. The same would true for a teacher who starts lessons with a new class for the first time. I think that recognising the "style" of AI-generated text is going to be an important skill, and it's one I've been trying to develop so I can apply it in my technical editing work.
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
Passage I
I settle on the couch ...
Passage II
Children and adults alike ...
Passage III
When we arrived there ...
OK, here goes. I think the first one may have had some AI input. That's because it's so bland and characterless, whereas the other two, especially Passage 2, have attitude and personality - something that is lacking in most AI-generated texts that I've seen.
Correct. So if we can spot/sense it, how hard would it be for a teacher who had been correcting essays for 30+ years? Also, you may notice that an average sentence length is much shorter in AI, with fewer complex sentences, fewer embedded clauses, etc. I took a couple of articles out of the AI text on purpose, just to create a few fake grammatical mistakes.
[Edited at 2023-04-14 12:45 GMT]
| | |
|
|
|
Ice Scream wrote: Philip Lees wrote:
I think that recognising the "style" of AI-generated text is going to be an important skill, 1. Surely the technology will improve faster than we can keep up? 2. If AI/MT is to be considered a suitable tool for translators, why not for students and everyone else? Isn’t this like barring them from using a computer?
My answers:
1. Eventually, probably, but I think it's going to be a while yet. One reason is that, as a given AI engine learns to simulate certain cognitive behaviours more effectively, in doing so it will inevitably develop its own "style", which I believe will be recognisable. Remember that these programs have no will, no volition. They are simply carrying out extremely sophisticated procedures of statistical analysis. Without human intervention, which at this scale would be impractical, statistical features of the source material would start to emerge in the finished product and could be identified by an attentive reader.
In the case of ChatGPT, in whose company I have spent an amount of time, this style could be characterised as a kind of blandness, underlaid with an almost pathological fear of giving offence (not really "fear", of course, but it's hard to avoid a bit of anthropomorphising). The bot also strives to avoid anything controversial. As a test, I asked it to write an op-ed piece explaining why abandoning the Good Friday agreement would not be bad for Northern Ireland. ChatGPT bluntly refused to do it, instead giving me a list of reasons why my proposed premise was not true. (I confess that this surprised me. I was not expecting such a response.)
This kind of bias, which must have been "hard-wired" in by ChatGPT's creators, will also be visible in the output.
So I think it will be possible, at least in the near future, to spot the bot, in the same way as I can tell easily whether a given guitar solo was played by Pat Metheny or Mark Knopfler, even if it might not be easy to explain precisely why. Of course, just by pure chance, the bot will sometimes get it "right enough", so this identification will not be infallible, but I think it's still worth trying to do it.
2. It depends on the task. If students want to use AI tools to gather, summarise and supplement their course material, I see nothing wrong with that. In fact, AI tutors will probably be a feature of the relatively near future. However, if students are specifically assigned to produce a piece of original writing, then using barely modified AI output would clearly be cheating.
If I'm wrong about the above points, then that means that, essentially, writing as a skill will cease to exist, just as mental arithmetic as a skill has ceased to exist since pocket calculators became widely available. I remember when that happened, as I was working in a tax office at that time and we were still doing calculations by hand, using complicated forms (like a spreadsheet, but made of paper!). Then the government issued each office with a single electronic calculator and our world changed.
Of course I can still do mental arithmetic. But how many of today's teenagers can multiply 12 by 13 in their heads? And is that a valuable skill, or one that humanity can let go of with a sigh of relief?
| | | | | I got it right! | Apr 15, 2023 |
Lingua 5B wrote:
Correct. So if we can spot/sense it, how hard would it be for a teacher who had been correcting essays for 30+ years? Also, you may notice that an average sentence length is much shorter in AI, with fewer complex sentences, fewer embedded clauses, etc. I took a couple of articles out of the AI text on purpose, just to create a few fake grammatical mistakes.
That kind of thing would be very easy to fake, so I wouldn't use it as a basis for identifying AI-generated text. With short examples like these, it's hard to be sure, but I think that with longer texts, or repeated texts coming from the same source, an experienced editor - and translators, of course, are also experienced editors - should be able to detect the dead hand of binary logic, even if wrapped in lavender.
| | | | | On the same subject... | Apr 15, 2023 |
I read yesterday a very interesting piece on a Portuguese weekly magazine (“Visão”) on the same subject. Here’s my translation, hope you like it (if you happen to spot any mistake, please keep in mind that my paid translations are only into my native language):
"(…) Today's Artificial Intelligence (AI) tools have thrown the rough fake news, still fresh in memory, into the land of child's play in a primary school playground. Now, facts are invented and go viral at the speed ... See more I read yesterday a very interesting piece on a Portuguese weekly magazine (“Visão”) on the same subject. Here’s my translation, hope you like it (if you happen to spot any mistake, please keep in mind that my paid translations are only into my native language):
"(…) Today's Artificial Intelligence (AI) tools have thrown the rough fake news, still fresh in memory, into the land of child's play in a primary school playground. Now, facts are invented and go viral at the speed of light (in 5G, obviously). Who, even for a second, didn't believe that the image of Pope Francis wearing a white padded jacket was real? Or who didn't consider as factual, even for a moment, the photo of Donald Trump walking, looking heroic, towards the New York courthouse, with a crowd of supporters following him?
This kind of image manipulation, which, until recently, was only within the reach of a chosen few experts, is, through AI tools, available to anyone - and accessible to help win debates. Current technology has reached a point where it forces us to constantly question what is true or false. There are already tools that can, in seconds, make someone say what they never said or “clone” a person into compromising situations as if they were real. With exhaustive manipulation within everyone's reach, any image is no longer proof of an event and even a speech can be altered so that an adversary appears to say the opposite of what was said.
In this reality, even reliable sources lose their... trustworthiness. In recent days, the editors of the British newspaper “The Guardian” discovered that, for example, a news article from their newspaper was being used as a source by ChatGPT, with that “news value” we readily embrace. Yet, in fact, that article never existed - it had been completely fabricated.
In a world where lies are becoming easier and more accessible, how do we go about uncovering what is true? These are the challenges all of us will be facing." ▲ Collapse
| | | |
Lingua 5B wrote:
They can add a few mistakes but can't change the entire flow and subtext. Impossible. We are talking about essays here, not translations.
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
Passage I
I settle on the couch with my coffee and book in hand. As I turn pages, the sound of rain outside creates a background melody that adds to the peaceful ambiance. It's a moment of pure bliss as I lose myself in the pages of book, forgetting about the outside world. I glance up from my book and see the raindrops sliding down the window, and I feel grateful for this moment of stillness.
Passage II
Children and adults alike need to experience how rewarding it is to work at the edge of their abilities. Resilience is the product of agency: knowing that what you do can make a difference. Many of us remember what playing team sports, singing in the school choir, or playing in the marching band meant to us, especially if we had coaches or directors who believed in us, pushed us to excel, and taught us we could be better than we taught was possible.
Passage III
When we arrived there, the house was empty. We tried calling them but nobody was answering the phone. So I took a stroll to the neighborhood to check if somebody could help us. There was an old lady down the road who was very friendly but said she didn't have contact with Mary and Steve and wouldn't know where they might be at that point. So we went to the village to have lunch and make plans for the rest of the day. We enjoyed some tasty local food in this lovely picturesque village.
[Edited at 2023-04-13 10:02 GMT]
I honestly don't know. I would say any of them could have been produced by AI. If I had to guess I'd say the last one was the human one.
But producing a good paragraph is not the same as producing a coherent text that makes sense as a whole and has some original thought behind it.
[Edited at 2023-04-15 09:00 GMT]
| | |
|
|
|
| Adding mistakes | Apr 15, 2023 |
Lieven Malaise wrote: Ice Scream wrote:
When I taught translation at university, the only work I received was actual translations. How would you tell whether these were AI-assisted? DeepL makes translation and interpretation errors and is stylistically far from flawless, but makes virtually no spelling and grammar mistakes, let alone typos. In Dutch (not the easiest language to write) that should be a huge red flag for any teacher dealing with inexperienced students. Any wicked student would of course add a few 'mistakes' to cover up for his godless behaviour.
That sounds like more effort than just doing the thing yourself in the first place.
| | | | Lingua 5B
Bosnia and Herzegovina
Local time: 13:35
Member (2009)
English to Croatian
+ ...
Rachel Waddington wrote: Lingua 5B wrote:
They can add a few mistakes but can't change the entire flow and subtext. Impossible. We are talking about essays here, not translations.
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
Passage I
I settle on the couch with my coffee and book in hand. As I turn pages, the sound of rain outside creates a background melody that adds to the peaceful ambiance. It's a moment of pure bliss as I lose myself in the pages of book, forgetting about the outside world. I glance up from my book and see the raindrops sliding down the window, and I feel grateful for this moment of stillness.
Passage II
Children and adults alike need to experience how rewarding it is to work at the edge of their abilities. Resilience is the product of agency: knowing that what you do can make a difference. Many of us remember what playing team sports, singing in the school choir, or playing in the marching band meant to us, especially if we had coaches or directors who believed in us, pushed us to excel, and taught us we could be better than we taught was possible.
Passage III
When we arrived there, the house was empty. We tried calling them but nobody was answering the phone. So I took a stroll to the neighborhood to check if somebody could help us. There was an old lady down the road who was very friendly but said she didn't have contact with Mary and Steve and wouldn't know where they might be at that point. So we went to the village to have lunch and make plans for the rest of the day. We enjoyed some tasty local food in this lovely picturesque village.
[Edited at 2023-04-13 10:02 GMT] I honestly don't know. I would say any of them could have been produced by AI. If I had to guess I'd say the last one was the human one. But producing a good paragraph is not the same as producing a coherent text that makes sense as a whole and has some original thought behind it. [Edited at 2023-04-15 09:00 GMT]
True, all of them had been plucked out of a bigger text/narrative, including the AI one. I asked the AI to produce a 1k word essay titled “Rainy Day”, then I pasted only one paragraph out of it here. Of course AI took it literally and started writing about physicalities of rain, while the best essays shouldn’t take the title literally, unless they are 8 year old writers.
| | | | Lingua 5B
Bosnia and Herzegovina
Local time: 13:35
Member (2009)
English to Croatian
+ ...
| Funny but true | Apr 15, 2023 |
Philip Lees wrote:
the dead hand of binary logic, even if wrapped in lavender.
So well put LOL
| | | | Lieven Malaise
Belgium
Local time: 13:35
Member (2020)
French to Dutch
+ ...
Rachel Waddington wrote:
That sounds like more effort than just doing the thing yourself in the first place.
But it was a joke, my point being that inexperienced students using DeepL would be easy to unmask.
| | |
|
|
|
Regarding chat, I can say that this is a very cool development that shows us in which direction handheld AI will develop, but I think it's too early to use it for translation, a little later when new algorithms are introduced, then we can count on something. I can say the same about writing research, doctoral, and qualification papers, for this I turn to my playful friends who do their job much better than chat GPT, but still the human side of the issue prevails over the AI side, because a perso... See more Regarding chat, I can say that this is a very cool development that shows us in which direction handheld AI will develop, but I think it's too early to use it for translation, a little later when new algorithms are introduced, then we can count on something. I can say the same about writing research, doctoral, and qualification papers, for this I turn to my playful friends who do their job much better than chat GPT, but still the human side of the issue prevails over the AI side, because a person can go deeper into the topic, emphasize what the AI will not even take into account. That's why I haven't fully delegated all the powers to chat GPT yet, and I don't advise you to do so either. ▲ Collapse
| | | | | I beg to differ... | Jul 23, 2023 |
Lingua 5B wrote: Ice Scream wrote: Philip Lees wrote:
I think that recognising the "style" of AI-generated text is going to be an important skill, 1. Surely the technology will improve faster than we can keep up? 2. If AI/MT is to be considered a suitable tool for translators, why not for students and everyone else? Isn’t this like barring them from using a computer? More like barring them from using their brain. Philip Lees wrote: Lingua 5B wrote:
They can add a few mistakes but can't change the entire flow and subtext. Impossible. We are talking about essays here, not translations. The only time I can imagine this being successful is in the case of new student, maybe transferred from another school, who submits AI-assisted work from the start, so you have no standard of comparison. The same would true for a teacher who starts lessons with a new class for the first time. I think that recognising the "style" of AI-generated text is going to be an important skill, and it's one I've been trying to develop so I can apply it in my technical editing work.
I will paste three passages below. Can you tell me whether some of them have been produced by GPT?
Passage I
I settle on the couch ...
Passage II
Children and adults alike ...
Passage III
When we arrived there ...
OK, here goes. I think the first one may have had some AI input. That's because it's so bland and characterless, whereas the other two, especially Passage 2, have attitude and personality - something that is lacking in most AI-generated texts that I've seen. Correct. So if we can spot/sense it, how hard would it be for a teacher who had been correcting essays for 30+ years? Also, you may notice that an average sentence length is much shorter in AI, with fewer complex sentences, fewer embedded clauses, etc. I took a couple of articles out of the AI text on purpose, just to create a few fake grammatical mistakes. [Edited at 2023-04-14 12:45 GMT]
You realize you can ask the AI "to make it spicier / edgier / give it more style", right?
If all fails, you can post-edit the text yourself.
That's not even mentioning that the AI field evolves very fast, and if OpenAI fails to offer something that gives the text more style and makes it less bland, SOMEONE will.
| | | | | Pages in topic: < [1 2] | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » ChatGPT and multilingualism | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
