Hello, dear Hivers 😽
In my opinion, ChatGPT is absolutely worthy of having a separate community here on Hive, and I would like to thank the moderator(s) for creating this community. I believe that it is an amazing tool, and we could have very informative and interesting discussions about it.
The purpose of this post is to encourage people to share what they are using ChatGPT for, discuss appropriate uses, and speculate about its future.
What I do use chatGPT for
I personally use ChatGPT to proofread or edit my English texts since it's my third language. Although I don't trust OpenAI completely, and having a good understanding of English is still required, it's a time-saver.
I also wish to use it for translation. For instance, one could take any eBook and translate it using ChatGPT, allowing you to read anything written in any language, which would be fantastic. However, it's not yet reliable, and I suspect it would make many logical and grammatical errors. Maybe in the future we will be able to do that.
Let's imagine that we reach to the a point when we can trust translations of ChatGPT 100% without double-checking it. The whole globe could communicate. But what if someone would publish translation as they own? Well I am guessing, it may not be a significant concern, since anyone will be able to use chatGPT to get perfect translation, and therefore, there will not be a market for it. On the other hand, professions such as editor or translator will be gone.
What I do not use chatGPT for
I strongly believe, that we should not try to create a content with AI. At least not yet. I know its a very strong claim, but I would suspect that there would be issues with plagiarism or content would be very “dry”. I don't say that is not possible in the future. I can imagine that A.I. will generate content as good as human (or maybe better) but maybe just in a very distant future.
But possibly it is already able to help with a structure or style of your content? I have not try it yet, but looks like its another topic worthy of another post.
I am not an expert on A.I. (OB) it is all my opinion and thoughts about it, but that's why we are here: to discuss and learn from each other. Let's share our knowledge, because that way we can accelerate our progress as individuals and as humankind as a whole. I look forward to read your posts or comments on how to put ChatGPT to good use.
Thank you for reading!
I don't think it's realistic to expect ChatGPT or any other AI language model to be 100% precise with translation. Probably 95% minimum accuracy would be more realistic. The United Nations translators are some of the "best of the best" and even they make mistakes.
I don't know if you are aware of this but people gathering to break down the Berlin Wall in 1989 was caused by a government official misspeaking at a press conference. Everyone involved spoke the same German language.
Of course if you can't understand the English language in the video talking about that you can copy-paste the video's transcript in ChatGPT.😁
There are things in language like idioms (a group of words having a meaning not deducible from those of the individual words) so certain details and concepts can be "lost in translation".
But I guess with a 98% or better accuracy ChatGPT could serve as a universal translator.
100% is probably too optimistic from my side... But do you think we are not able to "teach" A.I. to recognize idioms? What if it would have all information about our culture, history and language etc., would it be able to recognize idioms as well? or at least it would make remarks about all possible translations if it cant tell from the context? Your examples shows human error, because even smartest people can't know everything and we all do make mistakes. A.I. could have more data than any human ever could learn or am I too optimistic again? 🤔
I wouldn't say it's impossible. Even back in 2011 on the Jeopardy game show, IBM's AI supercomputer Watson provided the correct response to the clue, "A long, tiresome speech delivered by a frothy pie topping," with the answer "What is meringue harangue?" This demonstrates the capability of artificial intelligence to tackle complex language tasks.
keep in mind that the AI language model is based on human language which can have cultural and political biases and language can be either prescriptive or descriptive. Computers are great at recognizing patterns but imperfect humans created languages which don't conform 100% of the time to its own rules.
The Esperanto language was created to be as neutral as possible for all human speakers because of how diverse languages are.
Come to think of it, this goes back to the Chinese room thought experiment for A.I., doesn't it? It might seem like I was supporting that in my previous reply but I wasn't.
I see, thank you for your answer, you sure know a lot about A.I. Do you think, if we all one day would start using universal language (like Esperanto), at some point, it would change again, because language is "alive" new dialects and different meanings would emerge (for example American English and British English). Than it seems there is no hope to make absolutely perfect translations, which sounds very disappointing 😢
Going by the idea of the Kardashev scale I can imagine that if humanity reaches the point of being a globally unified Type I civilization then by necessity there would be one main language everyone would be using. Extending the idea of the Kardashev scale to humanity becoming a Type II civilization in which there are different colonies in the solar system maybe the same language would still be used. Going even further to Type III I would start to doubt whether humans spread throughout the Milky Way galaxy could maintain a unified language. If faster than light communication could ever be possible then maybe a unified language could still be maintained.
It's really hard to say without speculation. Right now there are countries like North Korea that use their own unique calendar. The international standard is the Gregorian calendar while some cultures still use a lunar calandar. It's hard for everyone to agree on a language when we can't even agree with an answer to "What is today's date?"
Would you like to volunteer to be a moderator in this community?
I would like to try, thought, I am not sure if I will be good at it :)
Done. You are now a moderator for this community.
Fictional content can be "dry" if you give it a simple prompt like "Generate a story about some kids going on a treasure hunt." I've found though that if you put a lot of details in the prompt about what you want in the story and request ChatGPT include sensory details in the prose the quality of the content is much better.
Because of OpenAI's content policy generating horror stories or a creepypasta about Jeff the Killer is virtually impossible without a jailbreak prompt. Even when using a jailbreak prompt sometimes the text stops being generated mid-way though the story and the text turns red with an error message.
ChatGPT gets really confused with certain fictional tropes like time travel and characters switching bodies. Hopefully in a few years ChatGPT can handle those kinds of higher cognitive details better.