Conversation with Perplexity (a dialogue about and with artificial intelligence)

This book not only addresses artificial intelligence as a theme, but also makes it an active part of the creative process. Instead of writing about it, the author chooses to write with it. Thus, this hybrid work is born, structured as a dialogue between person and machine.

Throughout its pages, Conversation with Perplexity unfolds as a series of dialogued encounters between the author and a real artificial intelligence application, which answers questions, argues, and even allows moments of humor, doubt, and reflection. The result is a reading that combines the rigor of thought with the constant surprise of an interlocution that is not entirely human, yet not completely alien either.

It is the first time—so far as is known in Catalan literature and possibly beyond—that a work starts from this premise: questioning artificial intelligence about itself. The book invites the reader to witness a kind of intellectual experiment where fundamental questions—what is consciousness? can a machine think? what role does creativity play?—cease to be mere speculations and become shared arguments between two voices of different natures.

From this encounter emerges a new way of thinking about the relationship between humans and technology. The dialogue format, carefully crafted and rich in nuance, allows the work to avoid both naive techno-enthusiasm and apocalyptic rejection. The author, true to his reflective style, keeps alive an intellectual tension between curiosity and skepticism, while the artificial intelligence—the Perplexity of the title—contributes its own logic, unexpectedly coherent and even poetic.

A work for our time

We live in an era in which artificial intelligence is no longer just science fiction, but a daily reality that influences the way we read, write, and think. Conversation with Perplexity arrives at the right moment to reconsider the foundations of this coexistence.

It is not a technical essay nor a philosophical treatise in the classical sense, but a living work that invites active reading and shared reflection. Its dialogic structure recalls the ancient Platonic dialogues, but with a contemporary irony: the disciple and the master are no longer two humans, but an organic mind and one of silicon.

The simplicity with which the author guides the conversation allows any reader, without needing technical knowledge, to delve into the complexity of the issues discussed. The language is clear and approachable, but the ideas presented are profound and may shake more than one conviction about what it means to think.

A literary novelty with a vocation for debate

The publication of this book is more than a cultural event: it is an invitation to debate the future of literary creation and thought. If, as the author suggests, an artificial intelligence can dialogue with us and contribute to generating new ideas, what does authorship really mean? Where do the boundaries between inspiration and programming lie?

The book "Conversation with Perplexity" is structured around a series of major themes that unfold naturally throughout the dialogue between the author and the AI. First, we find a sustained reflection on what exactly "artificial intelligence" means: whether it is legitimate to speak of intelligence when there is no consciousness, intentions, or deep understanding, or whether it would not be more honest to speak of a simulation of intelligence. From there, a second essential thematic block opens up, in which human intelligence and the functioning of current systems are compared without concessions: calculation speed, enormous energy consumption, and the use of trillions of parameters stand in contrast to the intuition, efficiency, and ability to get to the core of a question that are characteristic of human thought.

This first core naturally leads to a debate on language and the names we use. The author questions the term "artificial intelligence" and explores alternative proposals such as "excellent and extraordinary simulation of human intelligence" or "Computational Emulation of Intelligence", emphasizing that the name is not a mere technical detail but a key element in the expectations and misunderstandings that arise in society. In parallel, the book pays special attention to the issue of languages: how the quality of the answers varies depending on the language, what consequences this may have for speakers of minoritized languages, and to what extent this asymmetry can be read as a new form of linguistic discrimination. From this emerges the suggestive idea of an operative "internal language", a non-human working space where the knowledge of all languages is integrated without eroding their own richness.

Another fundamental axis is that of ethical and political risks. The dialogue dissects the dangers of attributing too much authority to AI: the risk that individuals, companies, or institutions may end up leaving serious decisions in the hands of automated systems, thereby de facto renouncing their moral responsibility. It raises the temptation to see the algorithm as a "neutral" and superior voice, and how this can lead, if not kept in check, to dehumanizing decisions disguised as technical rationality. The book also asks who should assume responsibility for potential harm: if AI cannot be responsible because it has neither will nor consciousness, then the focus shifts to designers, companies, and regulatory frameworks.

To all this is added a very concrete reflection on the functioning and limits of current systems: the absolute dependence on electricity and large data centers, the energy cost of training and using giant models, the lack of personal memory of conversations, and the impossibility of experiencing emotions such as loneliness or desire. The AI's answers, however convincing they may seem, appear as the result of a statistical process over large volumes of data, not as the expression of lived experience. This contrast gives rise to a discussion about the need to always maintain the distinction between simulation and reality, and to cultivate a critical attitude towards what the machine "says".

Finally, the book addresses two further strands that give it depth and warmth. On the one hand, the question of ethics and whether there can be intelligence without a moral foundation: what it means for a decision to be intelligent if it does not take into account its human consequences, and to what extent AI can or should carry some kind of ethical "criterion". On the other hand, the role of questions and of the author as interlocutor: it defends the idea that the AI's answers would not exist without the questions, and that the authorship of the text is, in a certain sense, shared. The conversation is therefore not only an interrogation of the machine, but also a mirror in which the reader can observe how they themselves ask, doubt, and think.

"Conversation with Perplexity" is, above all, a book of living conversation. Xavier Casassas Canals addresses an artificial intelligence application with a critical spirit, genuine curiosity, and a generous dose of intellectual provocation. He is not satisfied with superficial answers or technological slogans; he uses each answer to ask a new question, to refine a concept, to dismantle a cliché, or to expose an ambiguity. In this way, what could have been a mere curious experiment with a digital tool becomes a profound exploration of our time and of the way we think about technology.

The initial axis of the dialogue revolves around a seemingly simple question: is it legitimate to call a machine "intelligent" when it itself admits that it does not understand what it is saying? Using concrete examples and very clear comparisons, the author shows how current AI works on the basis of electrical operations, transistors, neural networks, and statistics, but has neither consciousness, nor intentions, nor an inner world. It can simulate a dialogue, it can produce texts that sound convincingly human, but it does so by recombining learned patterns, not by living or understanding. The book insists again and again that confusing this simulation with real intelligence is a conceptual error that can have serious practical consequences.

This idea becomes especially striking when the story of ELIZA enters the scene, the first chatbot which, in the 1960s, simulated a conversation with a psychotherapist. The people who interacted with it came to attribute understanding and empathy to it, even though the program only bounced back their own sentences with a bit of formal cleverness. Casassas recalls the surprise and anxiety of its creator, Joseph Weizenbaum, to show how inclined we are to humanize machines when they answer us with a certain tone and fluency. At a time when current systems far surpass that first experiment, the book invites us to ask what kinds of illusions we are willing to accept and what risks this credulity entails.

From there, the conversation opens up towards the terrain of ethics and responsibility. If an AI can give seemingly sensible advice about health, work, relationships, or political decisions, what happens when someone follows it blindly? Who is responsible for the outcome? The AI itself acknowledges that, without consciousness or will, it cannot assume moral responsibility; the author replies that, in that case, we must look to the designers, the companies, and the legal frameworks that make these tools available to everyone. The book puts its finger on the wound: it is all too easy to take refuge in the idea that "the machine said so" in order to avoid the obligation to critically examine what we do and decide.

This debate is linked to another delicate aspect: the way AI is presented in the media and in public discourse. On the one hand, there is no shortage of apocalyptic narratives that see in AI the prelude to the destruction of humanity. On the other hand, there is a technological euphoria that promises an almost paradisiacal life, free of work and full of comfort, thanks to algorithms. Casassas and Perplexity try to escape these two extremes and build a more serene perspective: AI is a powerful tool, with enormous potential for both good and harm, and what we do with it will depend above all on how we understand it and how we integrate it into our institutions, laws, and habits.

One of the most thought-provoking chapters is the one that revolves around languages. The author repeatedly asks the AI whether it answers just as well in Catalan as in English, and the machine eventually admits that it does not: the quality of the answers depends on the quantity and richness of the data available in each language. This opens up an uncomfortable debate: to what extent are we, without realizing it, creating a world in which there are first-class languages, which have access to finer and more complete answers, and second-class languages, condemned to poorer responses? The proposal of an operative "internal language" – not designed for humans, but as a shared working space where all knowledge is translated – appears as one possible way of facing this challenge. The book does not close the debate, but it does pose it with a clarity that invites continuation.

In parallel, the dialogue delves into the question of ethics as a condition of intelligence. Can we say that someone is intelligent if their capacity to reason and calculate incorporates no consideration of good and evil? What distinguishes a purely instrumental intelligence – able to find the most efficient means for any end – from what we would call wisdom? Through direct questions, the author forces the AI to acknowledge that current systems may imitate moral language, but they have no ethical foundation of their own; any "rules" they follow are imposed from outside. The reader discovers, almost without realizing it, the boundary between technical reason and moral responsibility.

Another guiding thread is the question as the engine of thought. The book shows how the quality and precision of the answers depend to a great extent on how the question is formulated, how much contextual information is included, and how far the interlocutor is willing to keep asking, to insist, and not to be satisfied with the first plausible answer. This idea culminates in an important claim: the answers generated by the AI are not only "its own", but the result of shared authorship; without the creativity of the person asking, there would be no text. Reading this conversation, the reader has the feeling of attending a mutual learning workshop, in which a human and a machine, each from their own place, gradually refine one another.

All this unfolds in a style that becomes increasingly careful and approachable as the book progresses. The author not only questions the AI about concepts, but also asks it to improve its writing, to avoid repetitive formulas, and to adopt a more fluid and humanly readable prose. The result is a text that combines conceptual rigor with an enjoyable reading experience, often full of images and comparisons that make it easier to understand technical issues without losing depth.

"Conversation with Perplexity" is not a book locked into a single thesis; it is, above all, an invitation. An invitation to distrust both uncritical enthusiasm and paralyzing fear, to look technology in the eye and to ask ourselves what the way we use it says about us. The reader who immerses themselves in it will not only learn more about how these tools work, but will probably come away wanting to ask many more questions of their own. And that is precisely what the book seeks: not to deliver the final word, but to kindle the desire to keep thinking.