Pages sur ce sujet:   [1 2] >
Using ChatGPT in Kudoz
Auteur du fil: Marjolein Snippe
Marjolein Snippe
Marjolein Snippe  Identity Verified
Pays-Bas
Local time: 01:52
Membre (2012)
anglais vers néerlandais
+ ...
Oct 26, 2023

I have a suspicion that one or more translators on this site are using ChatGPT (or a similar programme) to generate answers to Kudoz questions that they then present as their own answer. The answer sounds very confident, does not contain any references and can be complete nonsense, valid answers or anything in between.

I have seen answers given by ChatGPT, which have the same features, but in those cases it is clear that this is a machine-generated answer (as the user name is "ChatG
... See more
I have a suspicion that one or more translators on this site are using ChatGPT (or a similar programme) to generate answers to Kudoz questions that they then present as their own answer. The answer sounds very confident, does not contain any references and can be complete nonsense, valid answers or anything in between.

I have seen answers given by ChatGPT, which have the same features, but in those cases it is clear that this is a machine-generated answer (as the user name is "ChatGPT"). I would think it is undesirable that users present answers by ChatGPT as their own, as the answer can seem very convincing but is not backed by any understanding of the term. If the replier does not understand the term or the context to a sufficient degree, they will just paste in the machine's answer without editing. If they do understand the context, then surely they would not need a machine.

Would it be useful to add a tick box or something similar to have users declare that this is their own answer, not a machine-generated text?
Collapse


expressisverbis
Andriy Yasharov
Maria Teresa Borges de Almeida
writeaway
Jennifer Levey
Evandro Costa
Barbara Carrara
 
expressisverbis
expressisverbis
Portugal
Local time: 00:52
Membre (2015)
anglais vers portugais
+ ...
I've noticed something Oct 26, 2023

It's possible that this is happening.
Someone I've never seen around here before, perhaps a new member, has been giving a few answers, some of them totally senseless, and the tone of speech reminded me of answers generated by the bot.
I have no way of proving that, only my suspicions.
For now I think the only thing we can do is disagree with those members.

[Edited at 2023-10-26 12:19 GMT]


Evandro Costa
Barbara Carrara
Maria Teresa Borges de Almeida
Michele Fauble
Philip Lees
 
Joakim Braun
Joakim Braun  Identity Verified
Suède
Local time: 01:52
allemand vers suédois
+ ...
Darwin Oct 26, 2023

Marjolein Snippe wrote:

The answer sounds very confident, does not contain any references and can be complete nonsense, valid answers or anything in between.

High confidence level often accompanies low-quality answers.


Would it be useful to add a tick box or something similar to have users declare that this is their own answer, not a machine-generated text?


No, trust natural selection to take care of these Kudozers. (Depending on their rates, of course.)


Marjolein Snippe
Barbara Carrara
Maria Teresa Borges de Almeida
expressisverbis
TargamaT team
 
Jennifer Levey
Jennifer Levey  Identity Verified
Chili
Local time: 21:52
espagnol vers anglais
+ ...
Proz AI policy statement and action required Oct 26, 2023

There has been similar abuse of AI ‘tools’ here in the forums.
However, until such time as the Proz management issues a clear public statement governing the use, misuse, abuse (call it what you will … ‘infiltration’, perhaps?) of AI on this platform, we must expect to be subjected to a steadily-worsening ‘experience’ as we try to participate in our capacity as human beings.

To that end:
- The Kudoz rules – both for ‘asking’ and for ‘answering
... See more
There has been similar abuse of AI ‘tools’ here in the forums.
However, until such time as the Proz management issues a clear public statement governing the use, misuse, abuse (call it what you will … ‘infiltration’, perhaps?) of AI on this platform, we must expect to be subjected to a steadily-worsening ‘experience’ as we try to participate in our capacity as human beings.

To that end:
- The Kudoz rules – both for ‘asking’ and for ‘answering’ - need to be revised.
- The ChatGPT option needs to be eliminated from the Kudoz interface. See https://www.proz.com/topic/363811
- The Forum posting rules need to be revised to deal with the abuse highlighted in this present thread.

There are those amongst us, no doubt, who will say “well AI does no harm, so ‘let it be´”. Others will tell us yet again that it is sometimes difficult/impossible to know if AI has been used by a poster.
But the fact remains that while the use of AI in Kudoz and in the Forum is not mentioned in the relevant Rules, there is no formal basis for eliminating stuff that undermines the value of our contributions to these parts of the Proz platform.
JL
Collapse


Barbara Carrara
ibz
Maria Teresa Borges de Almeida
expressisverbis
Maria Rosa Fontana
Philip Lees
philgoddard
 
Zea_Mays
Zea_Mays  Identity Verified
Italie
Local time: 01:52
anglais vers allemand
+ ...
Funnily enough... Oct 26, 2023

Proz said that the introduction of Chatty in KudoZ is - apart from allowing translators to work at their best (sic!) - to reduce the number of these bot users... LOL.

Maria Teresa Borges de Almeida
expressisverbis
Philip Lees
Barbara Carrara
Maria Rosa Fontana
 
Maria Teresa Borges de Almeida
Maria Teresa Borges de Almeida  Identity Verified
Portugal
Local time: 00:52
Membre (2007)
anglais vers portugais
+ ...
@Marjolein Oct 26, 2023

You're not the only one with this suspicion! I wonder if it wasn't Proz itself, by including a ChatGPT feature, that opened the door, unintentionally, to these abuses…

expressisverbis
Philip Lees
ibz
Barbara Carrara
 
ibz
ibz  Identity Verified
Local time: 01:52
anglais vers allemand
+ ...
@Marjolein Oct 27, 2023

I doubt that you will get an answer. ChatGPT and ProZ seem to be BFF

Maria Teresa Borges de Almeida
expressisverbis
Barbara Carrara
Philip Lees
philgoddard
 
Marjolein Snippe
Marjolein Snippe  Identity Verified
Pays-Bas
Local time: 01:52
Membre (2012)
anglais vers néerlandais
+ ...
AUTEUR DU FIL
different Oct 27, 2023

But this is not about the site using ChatGPT. As long as they are transparent about using it, I don't mind. This is about one or more translators proposing answers, pretending that the answers are their own while in fact they have been copied and pasted straight out of ChatGPT.

Aurélien ARPAZ
Steven Capsuto
 
ibz
ibz  Identity Verified
Local time: 01:52
anglais vers allemand
+ ...
@Marjolein Oct 27, 2023

Yes, I know, but if ProZ gives ChatGPT the status of a member (with a very pompous profile) and allows the bot to answer questions, how can you then forbid other answerers to do the same as machine-generated answers obviously are "officially" accepted?

As you will notice, if you read the forums, this problem also occurs with translator contests.
And as I said before: ProZ is not willing to comment on questions regarding this problem. This, at least, is my experience. But may
... See more
Yes, I know, but if ProZ gives ChatGPT the status of a member (with a very pompous profile) and allows the bot to answer questions, how can you then forbid other answerers to do the same as machine-generated answers obviously are "officially" accepted?

As you will notice, if you read the forums, this problem also occurs with translator contests.
And as I said before: ProZ is not willing to comment on questions regarding this problem. This, at least, is my experience. But maybe you're lucky...!
Collapse


expressisverbis
Barbara Carrara
Maria Teresa Borges de Almeida
Maria Rosa Fontana
Philip Lees
 
Lieven Malaise
Lieven Malaise
Belgique
Local time: 01:52
Membre (2020)
français vers néerlandais
+ ...
. Oct 27, 2023

Marjolein Snippe wrote:
This is about one or more translators proposing answers, pretending that the answers are their own while in fact they have been copied and pasted straight out of ChatGPT.


What's the problem with this if the answer is provided with links to sources that back up the answer ? If it is not backed up by sources I would say that itself is the problem, not that the person answering the question used an AI bot. I would never blindly trust a non substantiated answer, even if it would come from a 30-year experienced professional. Not sure the use of ChatGPT itself is the main problem here.


expressisverbis
Barbara Carrara
Zea_Mays
Philip Lees
adelung
 
expressisverbis
expressisverbis
Portugal
Local time: 00:52
Membre (2015)
anglais vers portugais
+ ...
The problem... Oct 27, 2023

Lieven Malaise wrote:

What's the problem with this if the answer is provided with links to sources that back up the answer ? If it is not backed up by sources I would say that itself is the problem, not that the person answering the question used an AI bot. I would never blindly trust a non substantiated answer, even if it would come from a 30-year experienced professional. Not sure the use of ChatGPT itself is the main problem here.


That's precisely the problem.
Earlier this morning, I read several replies from a member who, to me, is very suspicious because of the "mechanical" answers given without providing any references.
He/She doesn't even indicate in his/her answer that used AI.
This member clearly acts like a robot. What's more, it's clear that the person is just starting out in translation...
In Kudoz, there are a few members I can trust and count on for their precious help. Answers without references/sources are rarely worth my attention.

[Edited at 2023-10-27 09:39 GMT]


Chris Says Bye
Maria Teresa Borges de Almeida
Zea_Mays
Philip Lees
 
Maria Teresa Borges de Almeida
Maria Teresa Borges de Almeida  Identity Verified
Portugal
Local time: 00:52
Membre (2007)
anglais vers portugais
+ ...
@Sandra Oct 27, 2023

expressisverbis wrote:

Lieven Malaise wrote:

What's the problem with this if the answer is provided with links to sources that back up the answer ? If it is not backed up by sources I would say that itself is the problem, not that the person answering the question used an AI bot. I would never blindly trust a non substantiated answer, even if it would come from a 30-year experienced professional. Not sure the use of ChatGPT itself is the main problem here.


That's precisely the problem.
Earlier this morning, I read several replies from a member who, to me, is very suspicious because of the "mechanical" answers given without providing any references.
He/She doesn't even indicate in his/her answer that used AI.
This member clearly acts like a robot. What's more, it's clear that the person is just starting out in translation...
In Kudoz, there are a few members I can trust and count on for their precious help. Answers without references/sources are rarely worth my attention.

[Edited at 2023-10-27 09:39 GMT]


On the other side, lately I've seen justifications so long that I'd swear they were ChatGPT or the like...


Evandro Costa
Chris Says Bye
expressisverbis
Philip Lees
 
Lieven Malaise
Lieven Malaise
Belgique
Local time: 01:52
Membre (2020)
français vers néerlandais
+ ...
. Oct 27, 2023

expressisverbis wrote:
Answers without references/sources are rarely worth my attention.


My point exactly. 😊


expressisverbis
Zea_Mays
Maria Teresa Borges de Almeida
Philip Lees
 
Marjolein Snippe
Marjolein Snippe  Identity Verified
Pays-Bas
Local time: 01:52
Membre (2012)
anglais vers néerlandais
+ ...
AUTEUR DU FIL
meh Oct 27, 2023

Some fair, level-headed replies. I suppose this is not a problem, it is just something I find annoying. If you use ChatGPT or any other machine, just say so. Fair point also about answers without references, so at least these people will get few, if any, points for their "efforts".
Rant over, back to more useful thinking!


Lieven Malaise
Aurélien ARPAZ
Maria Teresa Borges de Almeida
 
Maria G. Grassi, MA AITI
Maria G. Grassi, MA AITI  Identity Verified
Italie
Local time: 01:52
anglais vers italien
+ ...
I disagree Oct 27, 2023

I don’t see the need for KudoZ answerers to have to declare that they used AI.
In cases such as the latest gem I have come across in one of my language pairs, where the answer was about an exotic fruit when the question was about a chemical reaction, saying “I don’t speak a word of the language, asked AI, closed my eyes and hoped for the best” would be completely redundant.


Chris Says Bye
 
Pages sur ce sujet:   [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 »

Using ChatGPT in Kudoz






Trados Business Manager Lite
Create customer quotes and invoices from within Trados Studio

Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.

More info »
Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »