The role of artificial intelligence in digital publishing

It was already clear for some years that artificial intelligence would revolutionize many business and commercial realities, as well as society as a whole. However, the average web user perceived artificial intelligence as something unengaging, such as virtual assistants, while the idea of a digital entity capable of clearly understanding our requests and showing creativity was still considered abstract or even science fiction. In 2022, some tools initially considered curious and/or fun, like the generative artificial intelligence GPT-3, made a strong entry into many people’s lives. Millions today have realized that these “toys” were actually incredibly effective creative tools, out of the embryonic learning phase and ready to be used by different types of professionals.

In this article, we will focus on a business activity that has already undergone a profound revolution with the advent of digital technology and the web 2.0. The term “digital publishing” refers to the publication of books in the form of e-books, audiobooks, as well as online magazines and newspapers. It allows for the convenient publishing of books compared to traditional printing, as there are no costs related to physical production and distribution. Digital publishing has also allowed for the spread of self-publishing, where writers can publish their own books without going through traditional publishing houses. Additionally, technology allows for greater interactivity in digital books. For example, e-books can now include multimedia elements such as images, videos, and audio, and can also be enhanced with interactive features such as quizzes and games.
A digital reality that must be taken into account when discussing publishing is the blogosphere, which refers to the collection of all blogs on the internet. Blogs are websites that are frequently updated by one or more individuals and contain personal writings, comments, news, opinions, and other content. The blogosphere emerged in the late 1990s and rapidly developed, becoming an important part of online communication. Today, there are millions of blogs on a wide range of topics worldwide, from news to technology, cooking to relationships, politics to fashion.
There is a close relationship between the blogosphere and publishing as both are means of communication useful for sharing information and opinions, often influencing each other. For example, journalists and professional writers can use blogs to promote their work and interact with their readers. On the other hand, bloggers can use their platforms to bring attention to important issues and generate debates. Additionally, traditional publishing often uses the content from blogs for its own research and news-gathering activities.

Such evolution in the publishing industry has resulted in an interesting outcome over time, particularly for readers: the creation and spread on the internet of a large amount of articles, guides, essays, reviews in a completely free form. All publishing groups now have a basic offering, which can be defined as “entry-level,” where they offer various products for free. Of course, this does not eliminate profits, as in addition to advertising revenues, there are also marketing techniques designed to encourage the reader to purchase subscriptions, e-books, prints, etc. This is certainly very convenient, as free access allows anyone with an internet connection to access content that can serve to acquire basic (sometimes even in-depth) knowledge in almost any field. However, there are also negative effects such as a general decline in the quality of the material published, the proliferation of fake news and unfair practices such as clickbait. Furthermore, it has led to the majority of people today expecting to get everything for free. Twenty years ago, at the beginning of the 2000s, before the widespread spread of DSL connections and web 2.0, it was normal to go to a newsstand or bookstore to buy editorial products, today it’s not anymore, only a minority of people do it.

Therefore, today the problem arises of how to overcome the need to publish free material, essential for reaching a wide enough audience, without compromising the quality of the content. The creation of a well-written article, the result of in-depth knowledge on the topic, requires time and is possible thanks to the study and effort put in by the author. This is where artificial intelligence comes into play. With it, it is possible to create content in much shorter times while maintaining a quality higher than the standard that has been created in recent years, but this is only possible if a human figure who knows the topic well is guiding the AI. Furthermore, with little effort, it is possible to enhance the created material with “original” images, videos, and music, without neglecting the possibility of translating everything into different languages almost automatically. AI can be used to help correct drafts thanks to Natural Language Processing (NLP) techniques that are useful for analyzing text and detecting grammatical, syntactic, and spelling errors. In addition to grammar correction, AI can be used to identify potential semantic or content problems within the text. For example, AI can detect any inconsistencies or repetitions and identify coherence issues.

Based on my personal experience, I have begun conducting experiments to explore the advantages offered by currently available free AI. One of these experiments involves the translation of guides dedicated to well-known graphic software into English. In comparison to other popular automated translation solutions available on the web, how does ChatGPT perform in terms of translation quality?

What is the difference between your translations and those of a system like Google Translate?

“ChatGPT is a large language model trained on a vast amount of text data using deep learning techniques. Google Translate is a statistical machine translation system. The main difference between the two is the way they generate translations. ChatGPT generates translations based on patterns it has learned from the training data, it uses context and understanding of the language to generate translations, while Google Translate uses patterns it has learned from the translation data it has been trained on, to generate translations. As a result, ChatGPT is able to generate translations that are more human-like, and can handle more complex sentences and idiomatic expressions, while Google Translate may struggle with them.”

I can confidently say that ChatGPT was even modest in answering that question. Not only does its AI understand language, but it has at times demonstrated a deep understanding of the topics I have presented to it, providing additional information and insights in addition to the translation.

Another experiment I am currently working on is The AI Blog, an English language blog where I publish articles on technology, science, and art, with the headline stating: “All images and all text in this blog were created by artificial intelligences.” The idea of creating such a site came from reading the comments of people who are opposed to the proliferation of this type of technology (people fundamentally opposed to progress). It is intended to be a provocation, but it is useful for the topic being discussed in this article. Reading “All images and text in this blog were created by artificial intelligences” might initially lead anyone to think that it’s a site entirely created by AI. Of course, this would be a mistaken interpretation, as it is a blog created with AI and not by AI. The goal is to use artificial intelligences to create articles, but with human supervision to ensure that the content is consistent and based on the author’s knowledge of the topic. It is important to understand that a tool like ChatGPT is not perfect and can make errors in writing and answering questions. Those with a deep knowledge of a topic may be able to identify errors made by the chatbot. Therefore, it is important to use it with caution and not rely solely on its output. This should reassure those who fear being replaced or losing their job due to AI. Those who lack knowledge or expertise in a certain field will not be able to rely solely on AI to solve their problems. However, those who are open to using AI tools and are not prejudiced against them may discover new and interesting opportunities.

In recent weeks, there have been various rumors about a version 4 of ChatGPT with incredibly superior power compared to the current one, some claim that it would even be able to create an entire book of tens of thousands of words starting from a single prompt. However, the CEO of OpenAI has stated that these rumors are unfounded and there is no version 4 of ChatGPT under development or planned. For the moment, a ‘professional’ paid version is being experimented with, but it is not yet known how much it will cost or what additional features it will offer compared to the free version, other than hopefully greater server stability. Personally, I hope for the arrival of an official plugin to integrate the AI directly into WordPress (and maybe one day also in Adobe’s Indesign or Microsoft Office).

Therefore, we can say that artificial intelligences are already starting to play an important role in digital publishing. Here are a few ways they can be used now and in the near future:

  • Automatic writing: Artificial intelligences can be used to automatically generate text such as news, articles, and entertainment content. This can be useful for creating digital books as it can automate the writing and editing process and allow for greater writer productivity.
  • Content analysis: Artificial intelligences can be used to analyze the content of books and collect data on topics, style, tone, and other linguistic and writing elements. This data can be used to improve the quality of books and help writers better understand their readers.
  • Interactivity: Artificial intelligences can be used to create interactive digital books, where readers can make decisions that influence how the story develops.
  • Accessibility: Artificial intelligences can be used to create accessible digital books, such as voice books or books with enlarged text for the visually impaired or those with visual difficulties.
  • Artificial intelligences can be used to create graphic artwork, such as book covers or images for illustrated books or to enhance blog articles and websites. There are already some AI-based tools that can automatically generate images using techniques such as deep learning and the generative adversarial network (GAN).
  • In SEO, for example, AI can be used to analyze search data and optimize content for search engines. Additionally, AI can help identify and target the most effective keywords to increase website positioning in search results.
  • In copywriting, AI can be used to automatically generate the text for a sales email, advertisement, or landing page. Additionally, AI can be used to analyze customer demographic and behavioral data and customize the content based on that data.

Before we say goodbye, I leave you with a question that from now on could accompany you while reading any website: in your opinion, what percentage of this article’s text was created by an AI?