The AI Illusion: Why Machines Aren't Creative

by Luc JULIA

In The AI Illusion: Why Machines Aren't Creative, Luc Julia dismantles the hype surrounding generative AI by revealing what these technologies can actually do (as of today) versus what their promoters claim. Drawing on over thirty-five years' experience in the tech industry, Julia exposes the fundamental truth that generative AI doesn't create—it recombines existing data in response to prompts, producing impressive but ultimately derivative outputs that lack genuine creativity and understanding.

This essential guide takes listeners on a comprehensive journey through AI's past, present, and future, systematically debunking seven pervasive myths that shape public perception of artificial intelligence. Julia examines the technical limitations, societal implications, and environmental costs of generative AI while providing practical insights into how these tools function and where they're headed.

The book reveals the technical reality behind generative AI's "hallucinations," biases, and inability to reason or understand language; analyzes the economic and employment impacts of AI adoption across industries and society; demonstrates why artificial general intelligence (AGI) remains scientifically impossible with current approaches; and provides actionable solutions for more responsible AI development and regulation.

Genres: Technology, Artificial Intelligence

192 pages, Hardcover

First published March 16, 2026

Reviews

[Leo · Rating 4 out of 5]

Got the audiobook arc for review on netgally.

As someone who dont like the over use of AI and see many problematic things with it, I already knew a lot of points of the book. This can be a book to pick up to get familiar with the downsides with AI, what it is and its limitations. Think it gives a rather easy to folliw and tourough explanations and stories about it. 3.5 stars rounded up.

Side note: I got recommended by a friend to read a book about the positive sides about AI as I've only been focused on reading and watching about the negatives. Doubt it will change my mind but something I will look into in the future.

[suradha · Rating 2 out of 5]

This is a deeply confused book that has very little to do with understanding the basis of AI rather than what the author believes (generative) AI can and should do. because it starts off by advocating for a certain narrative of (gen) AI, which I presumed was talking to sci-comm enthusiasts but slowly slipped into advocating for the uptake and deregulation of the technology on the wider scale.

This is a book that tries to be almost self-help-y once you realise the author wants YOU, the reader to not be threatened by gen AI and the future of the economy and your dwindling agency and competency from the use of the various tools in the market currently, while couching it nicely into a framework of gen AI being unthinking and uncreative. Instead you, the AI prompter and the visionary robot whisperer are the primary agent of creation in the process.

This is a book that can only come by with a total lack of class consciousness, even if I think the author recognises most of the pitfalls of gen AI at this point. He simply cannot look past the institutional backing of AI and think about the proponents of the technology as being deeply financially embedded in the success of the technology- instead he frames any potential unused data centers as a public loss of millions of dollars, which simply isn't true (or doesn't have to be!). He wants to reckon with the environmental cost of the widespread use of the technology while he advocates for the widespread use of the technology in schools and at work when his solution to the energy and water crises are to limit unnecessary use of the tools personally. The fact that he cares at all about the environmental impact is likely because he's European, not because he's not an AI booster like any other.

We simply don't have time to get into the lack of evidence he presents for the future of AI being bold and beautiful, or that he believes AI cannot cause harm to humans unless the human explicitly asks for it to- there are several cases of AI assisted suicide and AI psychosis that would prove him deeply wrong at this stage. The fact that "a human" designed this tool to fail other "humans" does not mean the tool isn't to blame and that a human could cause the same scale of harm without the assistance of these deeply entrenched modern tools. The fact that he cannot weigh (gen) AI harms against one another proves to me that he is not writing for a general audience, but a set of boosters, boomers, or C-suiters who already believe in the inevitability of AI but are scared of Roko's basilisk or The Terminator. There is no in-between.

He also is quick to side with every tech mogul against European sanctions on data privacy and harm reduction policies on (gen) AI, citing that there are plenty of good uses of invasive data harvesting in medicine and security systems, neglecting the scale and the fact that any normal person on the street cannot opt out. Neither can he take into account that the terrorism he seems to want to deter was in fact caused by AI picking out targets in America's recent invasion on Iran and killed tens of children at Minab primary school, all because the humans operating it were deeply lazy and left any due diligence up to the (gen) AI search tool.

I thought this book would have a more broad scope understanding of the current state of AI given the stature of the author in the field, but turns out engineers are seldom taught to weigh the human impacts of their technology rationally and we all simply have to deal with the consequences of someone with no vision and no policy training releasing this book into the mass market.

Thanks Net Galley for the ALC of this book, it gave me a lot of perspective.

[mood_reading_maya · Rating 2 out of 5]

Thank you to RBMedia for the ALC.

Is artificial intelligence creative? How have large language and generative AI models impacted creativity? What impact has AI had on current workplace productivity? Are we in a hype bubble - and is it about to burst?

These are prevalent topics to take on when it comes to artificial intelligence. To their credit, Luc Julia does a fair job distinguishing between varying forms of AI - machine learning, expert systems, large language models, agentic models, etc. - in a way that is easy to understand. Indeed, when it comes to elaborating on tasks and processes where AI systems are successful, Julia breaks down limitations and successes with clear and approachable language and demonstrative examples.

However, I found this book and its supporting examples very anecdotal. One personal story or reported incident to support the authors argument or demonstrate a trend; it made The AI Illusion seem shallow. I want nonfiction that takes a stance and has the research rigor to back it up. This just didn’t! In the first half of the book, I wasn’t even certain whether the author was an AI proponent or adversary. If I can’t tell your stance 50% into the text, your arguments and conviction are lacking. Similarly, the ethical and environmental implications of AI development and usage were addressed with a light touch. Julia acknowledges they’re a problem then quickly moves on. For a book that purports to be solution oriented, the solutions were flimsy at best. The audiobook format made this feel like a long form podcast, which if that is your information format of choice then this will be a good fit for you.

Here’s the one bright side though, I think this book could be a useful entry point for white collar workers who have been forced to opt into AI usage out of fear of being “left behind.” A decent book club discussion for AI converts and skeptics. As for me…y’all can have your AI rapture and leave me behind, critical thinking intact.

[Joanna · Rating 4 out of 5]

I received an ARC of the audiobook from Netgalley for review.

The author obviously knows a lot about the technology and the creation of AI tools and has interesting anecdotes to share. The issue becomes one of definition. The author insists that AI tools aren't Creative because they are prompt-driven and its the prompts that allow for creation, not the AI demonstrating actually creativity. Similarly, he insists the tools are dangerous--it's the human users that are dangerous. The tool won't itself turn on humanity in a sci-fi dystopian way, but the human users might use its solutions to do dumb stuff.

These are important distinctions, but also easy to lose in the author's obvious cheerleading for the technology itself. This author believes AI is super helpful and that any problems are mistakes that can be fixed by better users and/or better data. He never really wrestles with the fact that these are near-impossible demands for a widely used tool.

Overall, I enjoyed the book, and particularly the clear descriptions of different types of AI (e.g., LLMs vs. expert systems).

This book is pretty technical for an audiobook. I think most readers would be better served reading this in print rather than trying to absorb all of the material in audio format.

[Desi · Rating 3 out of 5]

“We are becoming increasingly inventive to meet the demand of AIs even though they were originally created to serve us. We bend over backwards to ensure their existence while damaging our planet for little return.”

If I’m going to be a hater, I want to be an informed hater. I was skeptical to keep reading once I found out exactly who the author was, but research requires perspective from all sides of the topic. So I made myself keep listening. I was relieved to discover that Luc Julia seems to be of the same belief. Although his career is largely dependent on technological advancements and artificial intelligence, Julia discusses a fairly well-balanced amount of both the positives and negatives of generative AI.
The author spends a fraction of the book reassuring the reader that technology is far from being “intelligent” enough to develop a mind of its own and rebel against us Terminator-style. While this wasn’t a particular worry of mine, it’s still good to know. I’m moreso concerned about it ruining the world and humanity in other ways. The author does not shy away from discussing the ways AI data centers are harming the planet and how people are misusing the technology in dangerous ways.
Another important takeaway from this book, as the title suggests, is that AI cannot truly be creative. Julia tends to emphasize that, behind all the technology, humans are ultimately the ones doing the “creating”. They are the ones entering prompts and feeding AI its sources. Without humans, AI would not exist. It has no thoughts and is (mostly) only able to function within the parameters that its creator sets for it. It wouldn’t advance without us.
Julia calls for regulation on AI. Data centers should be limited to how much pollution they’re allowed to produce and how much energy can be used per facility. Theft of intellectual property is something that should be considered, too. He also touches on the notion that generative AI isn’t necessarily the best solution to every question. There are specialized AIs that have been trained for certain topics. Generative AI is not exactly designed to give users a CORRECT answer, just AN answer.

While there were a couple things I disagreed on throughout the book, one of the statements that the author makes in the conclusion is what stands out the most. Julia claims “[Generative AIs] serve to enhance human creativity. AI represents augmented intelligence, improving both human abilities and the functions of tools, allowing us to focus on higher-level intellectual and creative tasks.” I think that was the HOPE for AI. It was supposed to be a tool to assist us. While Julia seems to be of the mind that AI is not taking away as many jobs as we fear, I’d like to know more statistics on the current difficulties of being hired if you’re in an art career. Writing, graphic design, visual arts, animation, illustration, music. Businesses and organizations are very clearly skimping out on hiring real people to do these jobs and instead opting for poor-quality, cheap-looking/sounding AI-generated methods of acquiring the resources they need. Entrepreneurs are alright with looking tacky if it means they save a few bucks. Real people are in fact losing out on opportunities in their career fields because of AI. It is doing the jobs that people want to be and should be doing instead of the tasks that would actually make our lives easier. If AI had remained a tool rather than a toy, I might respect it more.

Additionally, I would have liked to hear more on the racism and classism that is prevalent in the placement of a lot of data centers. I understand that this is a broad topic, but at least one chapter to touch on the subject could have benefitted readers who are unfamiliar with that issue.

Overall, The AI Illusion is a decent book to pick up if you’d like to hear about it from someone experienced in the field. Luc Julia goes over the myths and facts that have spread along with the popularity of these technological advancements and gives you the information needed to develop your own opinion on them.


Thank you to NetGalley for allowing me a free audiobook copy of this title to review.

[Tilda · Rating 3 out of 5]

Thank you NetGalley and RBmedia for offering me a copy of this audiobook in exchange for an honest review!

This book is a perfectly serviceable introduction to contemporary discussions surrounding AI, particularly for readers who are not already deeply engaged with the field. Luc Julia does a good job of broadening the conversation beyond large language models and generative AI, something that is increasingly rare in public discourse at the moment. I liked that Julia explained that large language models are not the only types of AI and I echo his opinions that the complete pivot from expert systems to generative AI is a result of hype and we can only hope that this changes overtime.

The strongest sections of the book are easily those dealing with voice recognition and speech technologies. Those parts have a noticeably different energy to them and feel grounded in genuine expertise and not just surface-level research. I found myself wishing more of the book carried that same specificity and depth. The audiobook itself is also elevated considerably by the narrator, who gives an engaging and surprisingly dynamic performance despite the material being non-fiction and often technical. Honestly, the narration was probably the standout aspect of the experience for me.

That said, the book ultimately feels more broad than insightful. While it covers AI with reasonable breadth, very little of it felt new or particularly illuminating if you already follow developments in the field through work, academia, or personal interest. Much of the discussion felt repetitive, and some explanations - such as the brief mention of the Mandelbrot set - came across as overly reductive and surface-level.

I was also frustrated by some of the book’s arguments. The central thesis, that machines are not creative, is introduced early and reiterated at the end, yet surprisingly little time is spent actually developing or defending that position in a substantial way throughout the remainder of the book. Similarly, Julia’s argument that large language models should not be considered black boxes did not fully convince me. Ironically, his own explanation seemed to reinforce the opposite conclusion: LLMs are considered black boxes not just because laymen do not understand them, but because even experts often cannot fully interpret or trace the internal reasoning behind specific outputs in a transparent way. Something Julia does not mention and that I hardly ever see in non-academic discussions of AI is that we must remember that LLMs are not algorithms - there is a difference between machine learning networks and algorithms, and since LLMs are not algorithmic and trained on such a vast amount of data that their behaviour cannot be predicted even by the people who make them, they are black boxes.

I also found the discussion surrounding AI regulation somewhat disappointing. The book occasionally dismisses regulatory concerns too quickly, particularly when discussing AI surveillance systems. Using AI-powered cameras to identify terrorists cannot really be separated from the broader implications of mass surveillance and civilian profiling, which is precisely why many of these technologies face strong regulatory pushback in places like the European Union.

Overall, I would recommend this more to general readers looking for a broad and accessible entry point into AI conversations than to people already immersed in the field. It is not a bad book by any means, but it rarely moves beyond familiar territory or fully delivers on its most interesting ideas.

[WiseB · Rating 4 out of 5]

The book presents a sustained argument that contemporary artificial intelligence, including the latest generative AI systems, is fundamentally misunderstood by the public, the media, and many technology companies. The author contends that AI is a powerful engineering tool but not an intelligent entity in the human sense. Much of the excitement surrounding AI stems from exaggerated expectations, misleading terminology, and a tendency to anthropomorphize machines. It characterizes many claims about AI intelligence as illusions created by human interpretation of machine-generated results.

A section of the book is devoted to debunking what the author calls the "seven myths" of AI. These myths include beliefs that AI understands meaning, operates as an unknowable black box, will inevitably destroy humanity, is objective and unbiased, will eliminate most jobs, can learn anything, or can reliably make correct decisions. The author argues that each myth results from misunderstandings about how AI systems function.

The author also covers broader societal consequences of AI adoption … like environmental costs, excessive marketing hype, monopolistic tendencies, misinformation, malicious uses, and unrealistic business expectations. The author warns that many organizations engage in "AI washing," labeling products as AI-driven primarily for commercial reasons rather than because advanced AI capabilities are genuinely involved.

I like the author’s advocating a more pragmatic approach rather than pursuing visions of artificial general intelligence as an inevitable destination. AI should be viewed as a collection of specialized tools designed to augment human capabilities. Future progress should emphasize efficiency, sustainability, regulation, transparency, and responsible deployment rather than the pursuit of mythical machine intelligence. A realistic understanding of AI's strengths and limitations is therefore essential for society to benefit from the technology while avoiding its risks.

[Alison McIntyre · Rating 4 out of 5]

AI will not kill us or put us out of work. Phew that's a huge relief.

This was a well presented book about the pitfalls of AI and there sure are lots of them. He categorically lists everything that is currently wrong with AI and even proposed some solutions that aren't so farfetched. Towards the end he even said that how we will counteract the dissemination of false information would be that we develop critical thinking skills. Yay! Critical thinking is back baby.

For me, there were lots of 'aha' moments and 'I told you so' moments which I won't bother to repeat but the main point is that no one is fully thinking this through. It's a race for the egotistical tech companies to be the top but AI needs regulation and no one anywhere is mentioning or trying to regulate the biggest concern - the environmental impact.

Plus, I keep repeating this over and over but it's only as smart as the data it is programed with. Where does it get it's data? The internet and of course by stealing it. It then regurgitates the same over and over until it's sprouting incorrect garbage information.

AI can never innovate. It requires prompts to work and it needs humans to verify the information it presents. Plus the death of humans would mean the death of AI.

Thanks to NetGalley for the ARC of this book.

[Ebru Ekeman · Rating 3 out of 5]

A solid beginner's guide to AI from someone who has spent decades working in the field. Luc Julia does a good job of explaining the history of artificial intelligence, how generative AI actually works, and debunking many of the myths that dominate public discussion—from the idea that AI is genuinely creative to fears that it will inevitably take over the world.

If you've followed the AI conversation closely, there's not much here that will feel groundbreaking. The book is strongest as an accessible introduction rather than a source of new insights.

Ironically, I found the final chapters the most compelling. Julia explores the possible shift from today's general-purpose large language models to more specialized AI systems, the significant environmental costs of AI, and the need for responsible development and use, along with sensible regulation. These are some of the most thought-provoking—and often contentious—issues surrounding AI today. Unfortunately, they are also the least developed parts of the book. I would have happily traded some of the earlier material for a much deeper exploration of some of these ideas.

Overall, an informative and balanced introduction for readers who are new to AI, but those already familiar with the subject may come away wishing it went further.

[Katie · Rating 3 out of 5]

This is a case of where the message is clear, but the evidence is insubstantial.

The author has a storied career in AI. He's very good at providing simple definitions and explanations. He's very good at telling a tale, tracing the history, and making the story compelling. I tend to agree with his take on AI creativity. The problem is there is too much simplicity and very little support aside from claims, arguments, and anecdotes. I fear that this will not be a convincing take on the (generative) AI problem. At the same time, the author is so good at encapsulating the situation with concise and clear language ... anyone could understand, even if that understanding is shallow and in some cases insufficient. For example, reducing LLMs to lookup tables is far too simplistic a take.

The narration by Joshua Ward was superb. A text like this could read boring, but Ward elevates the source material and keeps it engaging.

Thank you to NetGalley and RB Media | Ascent Audio for the advance copy of the audiobook.

[Mona · Rating 3 out of 5]

Thanks to NetGalley and RBMedia for the advance audiobook copy.

Luc Julia has the authority for this subject, as he’s spent 35 years in the industry. So his breakdown of what generative AI claimed to be able to do vs what it can actually do, is well-handled. He argues that these tools remix existing content rather than create new content, so they aren’t really creative tools. It’s all written in an accessibly way, without any condescension.

Where the book struggles is in its evidence. Julia covers AI’s impact on creativity, employment, regulation, environment, but supports his argument with anecdotes rather than evidence. For a book that positions itself as myth-busting, it needed do better than “trust me bro”. While he is critical of how generative AI is being deployed, he still seems to be enthusiastic of where it is heading. That's not an impossible position to have, but I wanted a better why.

As an audiobook, it works well. The format suits the conversational tone and keeps the material moving.

[Curtis Mason · Rating 5 out of 5]

Know what you are getting. This book describes how AI will never be creative, nor will it ever take over the world. It also describes the need to regulate it appropriately and use it wisely. It is written by an AI engineer who has been working on AI for the last 30 years.

Does it accomplish the objective? Absolutely and in a great way. Luc does a fantastic job outlining the limitations of AI, its uses and the appropriate progression of AI in the future away from Large Language Models, LLMs, and towards smaller more directed and specific AIs. The author writes in a very approachable and understandable way without demeaning the reader.

As a data scientist working on a targeted AI, this was a breath of fresh air. It annoys me every time I read about some rich person warning of the AI apocalypse or when companies blame AI for the layoffs instead of citing the real reason, over hiring during Covid, as the culprit.

Writing: 5/5
Narration: 5/5
Overall grade: 5/5

[Sel · Rating 3 out of 5]

(Disclaimer: I received the audiobook from the publisher, but this has in no way influenced my review. My thoughts are entirely unbiased and honest.)

This is a good introduction book for those interested to know about Generative AI.

The narrator's performance really made the already-easy-to-understand explanations more accessible, especially for auditory leaners like me.

It's actually entertaining how he switches voices as the narrator, automated voice, or advertisement.

As for the overall arguments though, this non-fiction book is only 3 stars for because some topics could've been explored more and to be honest, it seemed that the author was also negating his earlier points.

More importantly, it's a bit odd that after establishing that capitalistic institutions are the reasons for the rampant and unregulated use of AI, the book concluded on a call-to-action for the readers.

It's also apparent that the lens of this author is still by a white man's.

[Dimitri · Rating 5 out of 5]

📕 Why (Not) to read this book (Target Audience)

Great book about realistic limitation off Ai.

👀 How this book changed my daily live (Takeaways)

The I in Ai was referred to information as in the abbreviation CIA

⁉ Spoiler Alerts (Highlights)

Generative Ai is not creative AI: it does not reason or think, its drive is to give an answer, in an 30% error rate (2025), it can only average what it already knows.

Creation and innovation are against the nature of AI, as innovation is about breaking rules and patterns. Ai follows the instructions and rules of its creator.

[Books Over People · Rating 4 out of 5]

I am pretty Anti-AI, so Julia was already singing my song. I appreciate all the data points specifically about the environmental horrors. Do we need to boil a lake so that you can make a crappy AI song or create some stupid image or to cheat on your homework? No. The answer is always no.

Thanks to NetGalley for the ARC.