| Pages in topic: [1 2 3] > | AI: Apparent consensus among freelancers who have tried it: It can be really useful. But be careful. Thread poster: Henry Dotterer
|
|---|
Hi folks. We have done a few surveys on the topic of AI. Many people have not tried it, but among those who have, there seems to be a fairly clear consensus. I would summarize the consensus as:
"Try it, it (can be/is) useful."
AND
"Be careful because it is sometimes wrong."
Over 80% of people who have tried it say it has helped them to be more productive.
Here are some comments:
Strong endorsement:
* You should use AI as an assistant.
* AI is quite resourceful for research purposes or for finding inspiration on alternative words/terms
* Using AI is a helping tool, makes translation more easier and faster. It also gives you inspiration and vocabulary expansion, gives you more choices.
* Try it, it can help you.
* Definitely do it. Master this tool.
Qualified endorsement:
* It is worth experimenting. The speed, relevance and quality of AI responses is astonishing❗
* I suggest that everybody ask any questions, clarify terms and check how close or correct the answer.
* I would suggest they dive in, they are bound to see a productivity boost as a result.
* Firstly, I'd say to get informed, trained and updated.
* Just go out there and try! Be creative
* If they can learn how to come up with proper prompts, which may be unique to your style of work, it will work well. You need to know how to use AI in order for it to be useful.
* I would say that we need to adapt to the market, and we need to get ourselves in the pool. Learn what it is about and try it out.
* Invest time to learn prompt engineering. Save prompts that worked well in a text expansion app for quick reuse. Keep honing your prompts over time.
Cautions:
* never trust ai and allways double check the translation yourself.
* You may experiment and use AI, but make sure you read and check every word of your translation, check consistency of terminology, and little details, since AI is still imperfect and you may find errors that are overlooked if not paying attention and could be significant, for example it can get confused with acronyms that are the same but mean different things for different fields.
* Make sure to come up with your translation first. Because once you see what AI came up with, you'll likely be pulled toward the AI generated translation.
* They should definitely apply, however, AI being an extensive field, myself being just a novice user (not having used premium features but just the free ones), can't foresee what a long-term, extensive use could bring.
* Learn how to communicate with AI. It is not an all-knowing resource. ChatGPT, for instance, is not always correct, so check information against other sources. Use it simply as another tool to help you with your work.
* Just ask the questions to AI one would ask a colleague, bearing in mind that explaining the context is very important. Bear in mind that the answer is not always correct, and double-check it. One may need to challenge the bot to get a better reply.
* Investigation is a key element in translation and people should use available tools to achieve a certain goal in a better and faster way. However, machines must be used as any other piece of equipment: human supervision and good sense are always essential.
Other:
* It depends on the client.
Comments welcome. "Concise" format. 
| | | | | Caution: confidentiality | Nov 2, 2023 |
Henry Dotterer wrote:
Cautions:
Thanks for the summing up. Among cautions we're missing confidentiality and intellectual property. For AI, I would stress this even more than with adaptable MT.
| | | | Metin Demirel 
Türkiye
Local time: 20:21
Member (2018)
Italian to Turkish
+ ...
Henry Dotterer wrote:
Over 80% of people who have tried it say it has helped them to be more productive.
How many people have participated in this survey? I think I missed the announcement for it. I would love to share my experience in detail if I had the chance.
| | | | | Please share Metin! | Nov 3, 2023 |
Metin Demirel wrote:
I would love to share my experience in detail if I had the chance.
Could you maybe give a summary here?
| | |
|
|
|
| Agree, Philippe | Nov 3, 2023 |
Philippe Locquet wrote:
Among cautions we're missing confidentiality and intellectual property. For AI, I would stress this even more than with adaptable MT.
Thanks, Philippe. I took "It depends on the client." to be getting at that.
| | | | | Number of participants | Nov 3, 2023 |
Henry Dotterer wrote: Metin Demirel wrote:
I would love to share my experience in detail if I had the chance. Could you maybe give a summary here?
Can you please answer his question. Thank you.
| | | | Zea_Mays
Italy
Local time: 19:21
Member (2009)
English to German
+ ...
| @Henry: pls change the format for sharing thoughts// New professional language service: | Nov 3, 2023 |
checking AI output.  ---- (Question: What is meant with "AI" here? DeepL is AI powered too.)
Seriously, I am adding this to my own services.
My experience with AI chat bots is: At first glance their answers sound very convincing and are obviously almost always perfect in terms of grammar or syntax. But at second glance quite often you spot completely wrong statements. When you tell the bot this but without exp... See more checking AI output. ---- (Question: What is meant with "AI" here? DeepL is AI powered too.)
Seriously, I am adding this to my own services.
My experience with AI chat bots is: At first glance their answers sound very convincing and are obviously almost always perfect in terms of grammar or syntax. But at second glance quite often you spot completely wrong statements. When you tell the bot this but without explaining what's wrong, it delivers new output with randomly changed elements.
Of course this often depends on the question and the prompt, but it happens that it gets simple math completely wrong.
[Bearbeitet am 2023-11-03 09:12 GMT] ▲ Collapse
| | | | Lieven Malaise
Belgium
Local time: 19:21
Member (2020)
French to Dutch
+ ...
I appreciate the effort to conduct a survey, but looking at your summary I'm under the impression that some points apply to ChatGPT and some rather to machine translation engines. I, for example, would say that (paid for) machine translation can be very useful for a translator, but ChatGPT and the like not so much (at least not for CAT tool users). So how relevant is it to draw general conclusions upon 2 different concepts, although both being AI based?
Furthermore I would also like... See more I appreciate the effort to conduct a survey, but looking at your summary I'm under the impression that some points apply to ChatGPT and some rather to machine translation engines. I, for example, would say that (paid for) machine translation can be very useful for a translator, but ChatGPT and the like not so much (at least not for CAT tool users). So how relevant is it to draw general conclusions upon 2 different concepts, although both being AI based?
Furthermore I would also like to know how many people participated. Making the sum of what I stated above and not knowing how many people participated, I can only conclude (at this point) that the survey results are pretty worthless and most likely biased. ▲ Collapse
| | |
|
|
|
ibz
Local time: 19:21
Member (2007)
English to German
+ ...
| Agree with Lieven | Nov 3, 2023 |
I don't see the relevance of this survey either, also because we have no detailed information about the basics of these statistics (questions, number of participants, etc.)
As to ChatGPT and translation, I think that everything has be said in another thread. AI in CAT tools is another subject.
As to one of the cautions cited in Henry's post:
[Quote]
* Just ask the questions to AI one would ask a colleague, bearing in mind that explaining the context is very import... See more I don't see the relevance of this survey either, also because we have no detailed information about the basics of these statistics (questions, number of participants, etc.)
As to ChatGPT and translation, I think that everything has be said in another thread. AI in CAT tools is another subject.
As to one of the cautions cited in Henry's post:
[Quote]
* Just ask the questions to AI one would ask a colleague, bearing in mind that explaining the context is very important. Bear in mind that the answer is not always correct, and double-check it. One may need to challenge the bot to get a better reply. [Unquote]
Sounds like training a wannabe translator to me - not something I would like to invest my time in ▲ Collapse
| | | | | The number of respondents is sufficient. I'm most curious about those who haven't tried it. | Nov 3, 2023 |
We ran several surveys and polls. It is hard to give an exact number of respondents because there may be overlap, and anyway I didn't tally them up, but the number of responses is in the hundreds at least, and it was enough that the picture that emerges "stabilized". The results are so much in agreement with each other, we can say with a good degree of confidence that most translators who try AI, find it useful but error-prone. Which may indicate that it is in fact useful but error-prone. No rev... See more We ran several surveys and polls. It is hard to give an exact number of respondents because there may be overlap, and anyway I didn't tally them up, but the number of responses is in the hundreds at least, and it was enough that the picture that emerges "stabilized". The results are so much in agreement with each other, we can say with a good degree of confidence that most translators who try AI, find it useful but error-prone. Which may indicate that it is in fact useful but error-prone. No revelation there; this just makes sense. (Of course, if you doubt these results, you don't have to demand answers from us, you can always run your own survey -- it is easy these days. Set it up on SurveyMonkey, which is free, and invite some translators that you know. If you get different results, please share!)
More interesting and perhaps thought-provoking than the above, in my opinion, is that our surveys suggest that perhaps half of respondents have not yet tried AI in connection with their work. I have less confidence in an exact figure on this one (due to self-selection bias, the fact that our pool may not be representative, a possible reluctance to answer one way or the other and perhaps other reasons.) But here, too, we can say with confidence, that there is still a significant number of translators who have not yet tried using AI in connection with their work.
To me, that is notable. Why not try it? Take some of your work, or if there is a confidentiality concern, some of your past work that has already been published, or somebody else's work, and ask ChatGPT, or Claude, or Bard or whatever, to translate it. Or ask it to comment on your own translation. See what you think. And maybe share your experience here. Thanks! ▲ Collapse
| | | | ibz
Local time: 19:21
Member (2007)
English to German
+ ...
| Surveys and statistics | Nov 3, 2023 |
I don't want to be tiresome, but can't you see that your conclusion is contradictory?
On the one hand, you say that enough people took part to be able to say that over 80 per cent find these tools useful.
On the other hand, you doubt the validity of the results about the people who don't use AI, namely
[quote] "due to self-selection bias, the fact that our pool may not be representative, a possible reluctance to answer one way or the other and perhaps other reasons" [unquote]
Why should this restriction only apply to this point and not to your first result (majority of respondents enthusiastic)?
And what's more and goes for every survey, the answers very much depend on the questions.
(There goes my second post )
| | | | Lieven Malaise
Belgium
Local time: 19:21
Member (2020)
French to Dutch
+ ...
Henry Dotterer wrote:
Of course, if you doubt these results, you don't have to demand answers from us, you can always run your own survey
Who exactly demands answers ? Giving answers is easy, by the way. Giving good answers isn't. As I said : I appreciate that you try, but this isn't the way to go. You can't even say how many respondents there were (and you ignore the problem that 'AI' can't be generalized from a translator's point of view), so how on earth should conclusions or discussions based on this survey have any worth whatsoever ?
Henry Dotterer wrote:
due to self-selection bias, the fact that our pool may not be representative, a possible reluctance to answer one way or the other and perhaps other reasons.
Glad you realise yourself that your survey is of little worth.
This is my second post, so I'm out I guess. We are forced to limit our reactions to a poor survey (which gives reason to disuss more elaborately) to 2 answers. I hope you understand how ridiculous and counterproductive this is.
| | |
|
|
|
| My concise opinion... | Nov 3, 2023 |
as far as translation is concerned, ChatGPT is not as good as DeepL, for example. Chatty is very good at rewriting stuff, but I would never trust it completely as a source for anything. That said, it can direct you towards the right solution. But, like MT engines, you have to check everything. So, it can be useful if taken with a massive pinch of salt...  ... See more as far as translation is concerned, ChatGPT is not as good as DeepL, for example. Chatty is very good at rewriting stuff, but I would never trust it completely as a source for anything. That said, it can direct you towards the right solution. But, like MT engines, you have to check everything. So, it can be useful if taken with a massive pinch of salt... ▲ Collapse
| | | | Metin Demirel
Türkiye
Local time: 20:21
Member (2018)
Italian to Turkish
+ ...
Henry Dotterer wrote:
More interesting and perhaps thought-provoking than the above, in my opinion, is that our surveys suggest that perhaps half of respondents have not yet tried AI in connection with their work.
80% of half the respondents is 40%. That means 40% of the respondents found it useful, and that is apparently not a consensus.
| | | | Kay Denney
France
Local time: 19:21
French to English
Giovanni Guarnieri MITI, MIL wrote:
Chatty is very good at rewriting stuff
Well, it is good at making sentences that sound like they make sense. If you need a bland style, sure.
My USP is my compelling creative flair, my rich vocabulary, my ability to come up with powerful puns and meaningful metaphors. I can produce translations that are far more interesting to read than anything Chatgpt comes up with.
I did try it once, just out of curiosity, and discarded all but three words of the drivel it produced.
| | | | | Pages in topic: [1 2 3] > | There is no moderator assigned specifically to this forum. To report site rules violations or get help, please contact site staff » AI: Apparent consensus among freelancers who have tried it: It can be really useful. But be careful. | Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
| | Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
