Effective Text Generation, Editing and Translation with ChatGPT

This article is an excerpt from the book, Modern Generative AI with ChatGPT and OpenAI Models, by Valentina Alto. This book will help harness the power of AI with innovative, real-world applications, and unprecedented productivity boosts, powered by the latest advancements in AI technology like ChatGPT and OpenAI.

In the world of natural language processing, ChatGPT stands as a powerful tool for various text-related tasks. From generating creative and coherent text to providing translations and editing assistance, ChatGPT offers a wide range of functionalities. In this article, we will explore how to harness the capabilities of ChatGPT to accomplish tasks such as generating engaging content, translating text between languages, and receiving helpful suggestions for editing. With practical examples and step-by-step instructions, we will unlock the potential of ChatGPT as a versatile text companion for developers and content creators alike.

As a language model, ChatGPT is particularly suited for generating text based on users’ instructions. For example, you   could ask ChatGPT to generate emails, drafts, or templates that target a specific audience:

Figure 1: Example of an email generated by ChatGPT

Another example might be asking ChatGPT to create a pitch structure for a presentation you have to prepare:
 

Figure 2 – Slideshow agenda and structure generated by ChatGPT

You can also generate blog posts or articles about trending topics this way. Here is an example:

Figure 3 – blog post with relevant tags and SEO  by ChatGPT

We can even get ChatGPT to reduce the size of the post to make it fit for a tweet. Here is how we can do this:

 Figure 4 – ChatGPT shrinks an article into a Twitter post

Finally, ChatGPT can also generate video or theatre scripts, including the scenography and the suggested editing. The following figure shows an example of a theatre dialog between a person and ChatGPT:

 Figure 5– Theatre dialog with scenography generated by ChatGPT

I only provided a truncated version to keep you in suspense regarding the ending…

Improving writing skills and translation

Sometimes, rather than generating new content, you might want to revisit an existing piece of text. It this be for style improvement purposes, audience changes, language translation, and so on.

Let’s look at some examples. Imagine that I drafted an email to invite a customer of mine to a webinar. I wrote two short sentences. Here, I want ChatGPT to improve the form and style of this email since the target audience will be executive-level:

Figure 6 – Example of an email revisited by ChatGPT to target an executive audience

Now, let’s ask the same thing but with a different target audience:

Figure 6 – Example of the same email with a different audience, generated by ChatGPT

ChatGPT can also give you some feedback about your writing style and structure.

Imagine, for example, that you wrote a script with scenography for your YouTube channel. You included the speech as well as images, clips, and video editing activities. You also know that your typical audience is between 15 and 25 years old. You want feedback on your script and ask for this from ChatGPT:

Figure 7 – Example of ChatGPT providing feedback on a video script

As you can see, not only was ChatGPT able to give me feedback about the writing style, but also it suggested how I could improve the scenography of the whole video, by including more visuals.

Again, imagine you wrote an introduction for an essay titled The History of Natural Language Processing and you want some feedback about the writing style and its consistency with the title:

 Figure 8 – Example of ChatGPT giving feedback on an introduction for an essay

Let’s also ask ChatGPT to make concrete examples of the attention-grabbing anecdote it talked about in its response:

Figure 9 – Example of ChatGPT elaborating on something it mentioned

I’m also interested in knowing whether my introduction was consistent with the title or whether I’m taking the wrong direction:

Figure 10 – ChatGPT provides feedback about the consistency of the introduction with the title

I was impressed by this last one. ChatGPT was smart enough to see that there was no specific mention of the history of NLP in my introduction. Nevertheless, it sets up the expectation about that topic to be treated later on. This means that ChatGPT also has expertise in terms of how an essay should be structured and it was very precise in applying its judgment, knowing that it was just an introduction.

It is also impressive to note how the model can give different feedback, depending on the context. With the video script, ChatGPT’s feedback took into account that the final consumption of that content would have been on screen. On the other hand, the essay’s introduction lives in a more formal and academic context, with a specific structure, that ChatGPT was able to capture once more.

Last but not least, ChatGPT is also an excellent tool for translation. It knows at least 95 languages (if you have doubts about whether your language is supported, you can always ask ChatGPT directly). Here, however, there is a consideration that might arise: what is the added value of ChatGPT for translation when we already have cutting-edge tools such as Google Translate?

To answer this question, we have to consider some key differentiators and how we can leverage ChatGPT’s embedded translations capabilities:

• ChatGPT can capture the intent. This means that you could also bypass the translation phase since it is something that ChatGPT can do in the backend. For example, if you write a prompt to produce a social media post in French, you could write that prompt in any language you want – ChatGPT will automatically detect it (without the need to specify it in advance) and understand your intent:

Figure 11 – Example of ChatGPT generating an output in a language that is different from the input

• ChatGPT can capture the more refined meaning of particular slang or idioms. This allows for a translation that is not literal so that it can preserve the underlying meaning. Namely, let’s consider the British expression It’s not my cup of tea, to indicate something that is not the type of thing you like. Let’s ask both ChatGPT and Google Translate to translate it into Italian:

Figure 12 – Comparison between ChatGPT and Google Translate while translating from English into Italian

As you can see, ChatGPT can provide several Italian idioms that are equivalent to the original one, also in their slang format. On the other hand, Google Translate performed a literal translation, leaving behind the real meaning of the idiom.

•  As with any other task, you can always provide context to ChatGPT. So, if you want your translation to have a specific slang or style, you can always specify it in the prompt. Or, even funnier, you can ask ChatGPT to translate your prompt with a sarcastic touch:

Figure 5.20 – Example of ChatGPT translating a prompt with a sarcastic touch.

The original content from: OpenAI’s Wikipedia page: https://it.wikipedia.org/wiki/OpenAI

Summary

In conclusion, ChatGPT is able not only to generate new text but also to manipulate existing material to tailor it to your needs. It has also proven to be very precise at translating between languages, also keeping the jargon and language-specific expressions intact.

Author Bio

Valentina Alto graduated in 2021 in Data Science. Since 2020 she has been working in Microsoft as Azure Solution Specialist and, since 2022, she focused on Data&AI workloads within the Manufacturing and Pharmaceutical industry. She has been working on customers’ projects closely with system integrators to deploy cloud architecture with a focus on datalake house and DWH, data integration and engineering, IoT and real-time analytics, Azure Machine Learning, Azure cognitive services (including Azure OpenAI Service), and PowerBI for dashboarding. She holds a BSc in Finance and an MSc degree in Data Science from Bocconi University, Milan, Italy. Since her academic journey she has been writing Tech articles about Statistics, Machine Learning, Deep Learning and AI on various publications. She has also written a book about the fundamentals of Machine Learning with Python. 

LinkedIn  Medium