As a Chinese, I clearly understand the impact of the language gap, but the birth of chatGPT makes people sometimes have to use academic language to make chatGPT understand what we mean, so why don’t we maintain this habit, so we Communication between people will also save a lot of unnecessary trouble. Although sometimes the expression of emotion may be weaker, but if we keep this habit for a long time, we will probably develop a universal language that will be used in the world. What a bright future!
Yes, I am totally for it
I am often very short with people, trying to leave out any sugar coating and excess words.
Not everyone likes that, and luckily we have amazing moderators @Ammon and @RealityMoez that can help add the needed context, or even explain that a short/to-the-point message from me is “programmer style communication” and not meant to be mean.
Some may argue that this very technical language is not “human communication”, and if needed I can identify as bot to make it easier for everyone
In some ways the advent of AI promises to (eventually) lead to easy, universal translation. Theoretically, that’s absolutely true. However, we instantly come to the difference between ‘theoretical’ and ‘practical’, in that pretty much all of the current AI LLMs are trained predominently on English language data and a bunch of other language dictionaries and translation related data.
The underlying language models are extremely heavily biased toward English language, partly because they are so often developed in countries where English is the predominent language, and where the staff who select the data sets for training mostly speak English.
That bias is further reinforced by the same thing that has limited Google’s decades-long use of Machine learning for its translation - that so much of the world’s available content is published in English, even by non-native English writers. There’s far less published in other languages, and in several cases, even if you took every single thing ever published in that language online, there wouldn’t be enough for machine learning to reliably learn from.
So, that’s the reason that when you read the documentation from OpenAI, or Bing, or Google, they all caution that their AIs only work fully in English, and on prompts in English. Any other language suffers a penalty to data performance, NLP, and thus to output quality.
Since I came into contact with GPT technology, I feel as if I have become a lot more honest. My own personality test is commander personality, and I have the ultimate pursuit of efficiency. GPT technology seems to be tailor-made for me, which is really great !
I have a deep understanding of this problem. It can be seen that the OpenAI official has been working hard to solve it, and the effect is very good. Even if I use some Chinese slang, chatGPT is still understandable and can even judge my emotions, but it is not enough The most important thing is that there are still flaws in the understanding of the context of non-English literary works. For example, when you ask a certain Chinese poet’s poem and ask chatGPT to answer the previous sentence of this poem, this little cutie is likely to make up a sentence that does not exist. It makes you feel very funny, although it does not complete the task you asked for, but it will have a unique sense of humor, a bit like Iron Man is telling Jarvis a nasty joke, but Jarvis still It’s just as fun to answer seriously.