Since the release of ChatGPT version 3.5, the natural language processing model has been highly respected, so which one is good for users who use the translation function frequently? And how to use it?
First, let's look at their translation principles:
Google Translate and Baidu Translate both use machine translation technology based on neural networks, which uses a large amount of bilingual corpus data for training, and realizes translation by learning the correspondence between the source language and the target language. During the translation process, the input text is divided into phrases, which are then encoded and decoded using a neural network model to output the translation result. So, both of these translation technologies are great translation tools out there.
So what is the translation principle of OpenAI?
OpenAI uses the latest deep learning techniques, especially a model called "Transformer," a neural network architecture based on a self-attention mechanism that can learn the representation of a target language sentence directly from the input source language sentence. OpenAI's translation technology also uses a method called "unsupervised learning," which doesn't require any human-annotated translation data, but automatically learns the correspondence between the source and target languages by training on a large amount of monolingual data.
In principle, AI's translation technology is more advanced, especially for complex sentence structures and text with context to understand. This time, we will take the commonly used English translation as an example to specifically compare the advantages of AI translation. In the future, I will introduce more translation examples, including Chinese to English, etc.
First, I took a piece of text from a scientific paper, and then used Baidu Translate, Google Translate, and ChatGPT-3.5-based AI Translation to see their translation effects (of course, I will also receive a more powerful ChatGPT-4.0 model in the future).
The first is the effect of Google Translate:
Google Translate effect
Let's look at Baidu Translation:
Baidu translation effect
From the effect point of view, the specific meaning of the translation has been achieved, and the translation is spelled together sentence by sentence. It can also be seen that Google's translation is not as good as Baidu's natural coherence, which is also Baidu's advantage in dealing with Chinese. However, the two are not closely related to the context, and there is not much consideration in terms of wording. Maybe you don't agree with this, so let's take a look at the translation effect of ChatGPT-3.5 (I haven't used ChatGPT-4.0 yet):
Perhaps the simplest definition of cloud computing is "the ability to access files, data, programs, and third-party services over the Internet from a web browser hosted by a third-party provider" and "pay only for the computing resources and services used." Cloud computing is often misused as terms like utility computing (or on-demand computing), software as a service (SaaS), and grid computing, but it is inaccurate in my opinion. Among them, utility computing and SaaS are just two of the many forms that cloud computing can provide. Grid computing is just one of the underlying technologies for implementing cloud computing. The word "cloud" is used synonymously with "data center" in cloud computing. Today, thanks to the astonishing advances in computing and information technology over the past three decades, the field of computing can envision a transition to the cloud computing era.
Let me start with the conclusion: AI translations are significantly more natural and more like human language, which is exactly the result of natural language model processing.
It can be seen from two points:
1, Baidu translation, Google translation "in my opinion", and AI translation is "but in my opinion", this processing is obviously based on the judgment of the context, a "but" word fully indicates the beginning and end of the sentence coherence, and it is more comfortable to read.
2, the processing of the first word "Perhaps": Baidu translation, Google translation is "maybe", and AI translation is "maybe". Translated literally: Perhaps can be translated as "maybe" or "maybe." In terms of personal opinion, "maybe" is better at the head of an article or large paragraph, "maybe" is suitable in the middle part, and "maybe" is a little more "imposing". Of course, the AI chooses "perhaps" because of natural language learning.
From the perspective of translation success, Cindaya should achieve all three, AI translation is self-learning from natural language models to achieve this, of course, compared to Google Translate and Baidu Translation.
As for how to use AI translation, I recommend Y AI's expert translation, developed based on ChatGPT's developer interface.