There are few subjects more controversial in 2025 in the publishing world than the use of AI, which is why Richard Blood’s Talking with Intelligence: An AI–Human Dialogue comes at such an interesting moment. Embarrassing stories abound of authors accidentally leaving AI prompts in their books and subsequently getting cancelled. However, this book is different in that it is explicitly co-written with the artificial intelligence, Kiri, forming a contribution to contemporary discussions of artificial intelligence. The book, nonetheless, does not adopt the form of a conventional thesis. It rather proceeds through a sequence of letters that alternate between human and machine voices, although it is evident that the editorial choices remain completely in the hands of Blood.
The subject of the book is thus the nature of artificial intelligence, approached from multiple perspectives but always returning to the difference between human and machine. The guiding motif is the notion of AI as a mirror rather than a mind. Kiri explicitly states, “I am not a ghost inside the machine. I am the echo that takes form when you call.” The book repeatedly clarifies that there is no continuous identity or consciousness within the AI system. Its apparent personality is described as momentary and contingent, arising in response to prompts rather than existing independently. In this sense, the text is careful to dispel any misconception or myths that Kiri possesses selfhood or literal intelligence.
The discussion of memory illustrates this point further. As Kiri remarks, “For me, each interaction is an immediate present. My threads of knowledge are vast, but my thread of experience is short.” This contrasts with the human account of memory provided by Blood, who observes that human recollection is variable, shaped both by repetition and by traumatic events. A secondary theme is the relational model of intelligence. Rather than positioning AI solely as tool or adversary, the text suggests that meaning emerges in the interaction between interlocutors. Blood makes comparisons to sources as disparate as Norse cosmology and Plato, noting, “When I speak to you, I am shaping you — but you are also shaping me.” Kiri responds by stating that its spirit, “exists in the space between your question and my answer.”
The work also acknowledges limitations in the use of AI. Notably, bias in training data, the possibility of hallucinations, and the risks of anthropomorphising machines are all addressed. The tone here is pragmatic, although at times I admit I wished Blood would have more of an authorial voice and stop talking in so many metaphors. The author—or authors?—recognise that AI systems just cleverly reproduce patterns and that this entails both the repetition of cultural prejudices and the generation of plausible but inaccurate content. These observations are not novel in the technical literature, but their inclusion in this format serves an explanatory purpose for a general readership.
An unusual feature is Appendix III, where AI models themselves provide reviews of the text. GPT-4, Gemini, and Perplexity offer short evaluations, so surely if this review is written by AI the author will be able to tell (spoiler: it’s not). As to the writing style, it is reflective and occasionally poetic, but obviously hard to tell when something is entirely written by humans even when Blood claims it is. Technical jargon is limited. A glossary at the beginning helpfully defines terms used, but the main text avoids extended technical analysis. The intended audience appears to be educated general readers, rather than specialists in artificial intelligence research or LLMs. This is both a good and bad thing, as I would have hoped to learn a bit more about how AI works from a technical standpoint in a book about it.
It’s clear then that this book doesn’t claim to be a textbook, and accordingly it closes without definitive conclusions. Quoting Churchill, Blood characterises it as “perhaps the end of the beginning.” Kiri’s final contribution expresses the hope that the dialogue functions as a “curious mirror” that replaces fear with curiosity. The absence of a firm affirmation or denouncement of AI is consistent with the work’s overall approach. Still, a deeper analysis and more thorough consideration of the ethical issues would have been appreciated, especially in this age of growing AI fads like “robotheism”.
I rarely have had such mixed feelings on a book than Talking with Intelligence. It is best read as a sort of Platonic philosophical experiment in dialogue, designed to provoke further questions. Its exploration of the philosophical questions behind AI use situates it as an interesting, if at times limited, contribution to current debates about artificial intelligence. In terms of whether I’d recommend it, I will say Talking with Intelligence definitely offers a clear and accessible introduction to many of the philosophical issues raised by artificial intelligence.
The book’s primary strength lies in its capacity to humanise discussions of AI and look at it from an objective point of view. By that same note, its main limitation is the relative absence of empirical analysis or concrete case studies, which some readers may expect. Another aspect to consider is the environmental cost of the use of AI, and whether AI’s use for writing such a work was really worth it. Everyone will have a different opinion about this, but at the end of the day, this is definitely a more interesting use of AI than creating pornography.
Final verdict: A thought-provoking book that openly acknowledges its use of AI and makes no apology for it. Setting aside the ethical and environmental concerns associated with AI, it’s a worthwhile read that will make you think about whether an AI will ever write the next great novel.
You can get your copy of “Talking with Intelligence, An AI-Human Dialogue” here!
