Introduction¶

The arrival, or rather the explosion, of the phenomenon of generative artificial intelligences for images (those I deal with in this text and which from now on I will refer to as AI for convenience) has brought with it a true revolution in the creative world. This technological innovation raises several questions, some of which are completely new in the artistic and technological landscape. Here I try to answer two of these: the first is who is the author of a work generated with the help of an AI, the second is what is a work.

In the first part of this text, I offer a personal reflection on what it means to be an author in the era of generative AI for images. I start from my experience as an author, developed well before the arrival of these technologies, and try to integrate what I have understood about the mechanisms of these new tools. The goal is to offer as general an analysis as possible, not tied to any specific AI, but one that can provide a complete overview of the challenges and opportunities that these technologies present for the concept of authorship.

A common element of all generative AI, which is important to emphasize, is that none generates images without human input. This confirms the necessity of human presence in the creative process, a fundamental aspect when discussing authorship in this new technological context. The relationship between human input and AI output thus becomes a central point of our analysis.

In the second part of the text, I try to describe what court rulings and regulatory texts say today, without any pretension of doing the work of legislators, lawyers, and judges. I try to describe some judicial cases collected literally from every part of the world to understand if a common thread can be found. Spoiler: this common thread seems to exist.

This discussion inevitably focuses on the concepts of copyright and author's rights, well-established notions that I believe must nevertheless be reinterpreted in light of new technologies. And this leads to new challenges with future implications for artists, creatives, and professionals in the field.

Who the author is constitutes THE fundamental question in this debate. It arises from the fact that, for the first time in history, we are facing a creative tool that does not faithfully execute the request of the human artist. This represents a paradigmatic change in the creative process, a change that deserves careful consideration, also to understand how the arrival of these machines is changing the creative process.

A phrase by John Culkin, a collaborator of Marshall McLuhan, describes so well what is happening that it could even be the subtitle of this book.

We shape our tools and thereafter our tools shape us [See John Culkin, A schoolman's guide to Marshall McLuhan in Saturday Review, March 18, 1967]

Traditionally, the tools used in the world of images faithfully transformed our inputs. Brushes for painters translated hand movement into marks on canvas. Similarly, software like Photoshop (in its pre-AI version) transformed our actions on the various components of the interface into graphic marks on the screen. In all these cases, the result was predictable once the tool used was learned. Similarly, an expert painter knows exactly what type of stroke a particular brush will produce on a specific canvas.

Now, for the first time, we have a tool that responds to our inputs, but whose exact response we cannot be certain of predicting.

This introduces a new level of complexity in defining authorship, as the final result is no longer exclusively determined by the artist's actions. The AI becomes a kind of collaborator in the creative process, introducing an element of unpredictability and, potentially, its own creativity.

The degree of uncertainty in the AI's response is variable, depending on the specific tool we are using and, above all, how we are using it.

One of the causes is the level of indeterminacy inherent in the use of the prompt, that is, the request—generally written—that the person makes to the machine describing what they want to obtain. To better illustrate this concept, let's consider a concrete example: faced with a generic request, such as "a car," as a response I could have an image of a Ferrari or a Fiat Panda. The AI has a wide range of possibilities to draw from, and the result could be any type of vehicle that falls into the "car" category.

If I start to specify my prompt better, for example asking for "a red sports car on a track," I begin to have more defined results. However, I will always have a part of the image not described by the prompt. The AI will intervene on elements not specified by the user such as the details of the car, the appearance of the track, the weather conditions, or even add unspecified background elements.

But a certain degree of indeterminacy will also exist for what we consider our subject; it is in fact impossible to describe it completely using words, and so even returning to the car example, the AI will intervene, in its own way, to complete the details of the vehicle not defined by our words.

There is one last element that contributes to the uncertainty in the machine's response, due to the mathematical method used to actually generate the images: the use of "chaos."

In short, it can be said that unpredictability in image rendering is a constitutive characteristic of the system, as an intrinsic attribute of the system.

This variability may be more or less high depending on the prompt. Here I limit myself to the textual prompt, but it's important to note that a prompt can be composed of a textual part, parameters, and reference images. This set can be called an "extended prompt," and includes a variety of different types of input.

It is in this variability and interpretation that the AI shows its capacity to be potentially creative. It does not limit itself to faithfully generating what is asked of it, but interprets, elaborates, and sometimes "invents," showing a form of computational creativity that challenges our traditional conceptions of artistic creation and that the person interacting must know how to grasp and direct.

This characteristic of generative AI makes them unique in the history of creativity, opening new possibilities but also new challenges in defining what it means to be an author in the 21st century.

This aspect makes working with a generative AI something truly new in the history of creativity. The artist is no longer the sole direct author of the work but becomes a kind of director who guides the AI through prompts and selections, in a collaborative human-machine creative process.

A new dynamic thus presents itself that raises fundamental questions about the nature of authorship and creativity itself.

Who is truly the author in this process? How much does human input count compared to AI processing? How is creative contribution evaluated in a work generated with artificial intelligence? How should the AI user manage these outputs? These questions are not only theoretical but have significant practical implications in the field of copyright, intellectual property, and in determining the creative value of works.

These are the topics I will discuss in the next chapters, seeking to better understand the emerging new creative paradigm and its implications for the world of art, copyright, and creativity in general. To do this I will present case studies and examine the current orientations of the legal world, while also trying to outline possible future scenarios for authorship in the era of artificial intelligence, also discussing the concept of work, which perhaps also needs to be revised.

This journey through the new landscape of AI-assisted creativity will lead us to reconsider not only what it means to be an author, but also what it means to be creative in an era when the line between human and machine action becomes increasingly blurred. Get ready to explore a fascinating and rapidly evolving territory, where art, technology, and human identity intertwine in ways never seen before.

Throughout the book, I will refer to both the concept of "copyright" and that of "author's rights." These terms, although often used as synonyms, actually have slightly different applications in various legal systems. For simplicity and fluidity of the text, I will sometimes use only the term "copyright" to refer to both concepts, bearing in mind that the specific applicability depends on the legal context of the specific country. This choice is purely practical and does not intend to ignore the differences between the two terms.

Acknowledgments¶

The idea of writing a book often arises from the conviction of having something significant to share on a topic; however, it is only through the interest of others that this conviction finds confirmation. In the case of this book, I was fortunate to meet people who appreciated both the theme and the way I approached it.

I wish to thank:

• Nicola Cavalli of Ledizioni publishing house, for believing in this project and welcoming it into their catalog.
• Simone Aliprandi, series editor, for his valuable advice and for actively contributing with some of his reflections.
• Athos Boncompagni, for his passionate preface.

The support of these professionals not only confirmed the timeliness of the topic discussed but also contributed significantly to the realization of the book.

A note on methodology¶

A final important note: to create this text, I integrated artificial intelligence into my revision process. My methodological approach developed organically and in a structured manner. I always started with a first personal draft, a conscious choice to ensure that the content was authentically mine. Subsequently, I created a personal assistant using Claude, specifically configured for text revision, which assisted me in the subsequent phases of work. The revision took place chapter by chapter, in close collaboration with this digital assistant, always maintaining creative control of the process.

Using this approach brought several benefits to the writing process. First, considerable time savings, allowing me to concentrate on other crucial aspects of the work. At the same time, the overall quality of the text saw a significant improvement, thanks to the assistant's ability to identify and suggest corrections for subtle errors that might escape a traditional review. Perhaps the most interesting aspect was the increase in creative and decision-making control: the assistant provided suggestions and alternatives, but the final decision on every change remained firmly in my hands, allowing me to refine the text according to my writing style and in general according to my authorial vision.

I feel I can say that this method transformed my writing process, making it more efficient without compromising my being an author.