ChatGPT vs Google Translate
Have we ever found ourselves reading something online that is written in, say Spanish, and then need to translate that to English?
Or receiving a WhatsApp message that is in French which we don’t understand and we need to translate it to English? Or an email message that needs to be translated?
We can translate any text from one language to another using Google Translate. It supports more than 100 languages, and can recognize speech as well as images — and even handwritten text.
Google Translate can translate text on the fly while we are online. We can also install the Google Translate app on desktops as well as mobile devices. It is one of the most widely used translation apps.
However, the technology behind Google Translate is such that it mainly translates word for word. In this process, it is not able to recognize the “meaning” of the word within the context of the sentence. This puts the accuracy of its translation into question.
Here comes ChatGPT. This revolutionary deep learning model is able to “meaningfully” translate text because its pre-training process has allowed it to understand better how sentences are constructed in different languages.
ChatGPT, like other machine learning models, produces much more accurate translations based on context and intended meaning. This is the right direction moving forward as Google has also augmented its Google Translate to incorporate some AI-driven features.
As an example, the English language uses idioms frequently to convey intent figuratively. A non-AI powered translation engine will not be able to recognize that the text contains words that don’t literally mean what it says.
Take this idiom:
“Kick the bucket”
The meaning and intent is to die or break down or stop working.
But if taken word for word for its meaning, it is saying to go and physically kick the bucket.
Here is another example:
Son: Dad, I hate Math. I don’t like problem solving.
Dad: Son, your Math exam is tomorrow. Can you just bite the bullet and burn the midnight oil just for one day?
The good news is when I try to get Google Translate to translate the above, Google is able to recognize that there is an idiom and translate it to the correct literal words.
Suffice to say that Google Translate does have some AI capabilities incorporated into it.
ChatGPT has acquired the knowledge of the intended meaning of idioms through its pre-training and deep learning and is thus able to translate it more accurately. It is also able to translate based on the intended meaning of the whole passage presented to it.
An AI-trained model depends heavily on its knowledge base in order to be able to translate meaningfully.
So if there is not enough content on the web on a particular language, say the Cambodian Khmer language, the AI model will never be able to learn and to pre-train itself on this language.
Obviously this will also mean that it will not be able to do a good translation job for that language.
An AI language model will thus be able to translate text from English to Spanish much more accurately than to a less commonly used language like Greek.
On the business side, Google is already monetizing on its Google search platform through Google Ads while OpenAI’s ChatGPT does not have any similar monetization model.
However, ChatGPT does have a subscription-based business model called ChatGPT Plus. At a cost of $20 per month, ChatGPT Plus ensures availability during peak times and with faster response times. Plus priority to new features and improvements.
In AI technologies that use the deep learning model, the brainchild is only as good as the knowledge that it has been trained on.
Content creators like us are the real ones that power the AI model. Without our content, where does ChatGPT learn from?
Cheers everyone!
I love what writing does to me — it reinforces what I already know.
