Can you use AI in book writing? And why the answer is both YES and NO.

Article summary:

AI can be an incredibly powerful tool for all authors, including book writing. It’s great for brainstorming, structuring, gap filling or breaking through writer’s block. But it’s not a substitute for your voice. And frankly, AI still struggles with long form content in many areas, including readability, reasoning and originality. The key in your own book writing? Use is as an assistant, not an author.

Unlike most writers, I’m a huge fan of AI. In fact, one of my favourite jobs ever, was working with an author on her book about AI. (If you haven’t read AI Magic, I’d highly recommend it even if you’re not a professional services firm. There are so many excellent lessons in it about how to adopt AI in a way that helps and doesn’t hurt your business!) And I love using voice chat while I’m driving to talk about brainstorm ideas.

But what about when it comes to writing books? There’s not a lot of research out there about how good AI is at writing books. If you do a search, you’ll mostly find writers giving you articles about why it’s the worst (I wonder why) and tech expertstalking about how good it’s getting (again, I wonder why). And people who want to sell you something taking one side or the other. Not that helpful.

So, I want to give you my honest opinion as a book writer who loves AI and who has had a massive exposure to AI content in the dozens of drafts of manuscripts that have come to me in the last year.

And the answer is simple. YES – you can use AI in book writing – and NO – you can’t.

Let me explain.

Can you use AI in book writing?

AI is a fantastic tool. World changing in fact. It’s such an advancement it’s like it’s going from stone tablets to smart phones.

If you’re reading this and shaking your head, it’s time to get on board. AI is not going away. It is going to change the way we work (and write) and if you don’t get on board you are absolutely, without a doubt going to be left behind. Yes, I said it.

BUT it’s not ready for all writing tasks. So, here’s when you can and should use it and when you can’t and shouldn’t when it comes to writing your books.

When the answer is YES

1. Brainstorming 

General purpose AI tools like ChatGPT are excellent at helping you brainstorm new ideas or expand on ideas that you already have.

If you’re a mortgage broker, you might ask it to suggest some of the top pain points that buyers have in today’s market. If you’re a leadership expert you might ask it to help you pinpoint a segment of an audience that is underserved in your industry.

It is great at this. And you can have an easy back and forth conversation where you continuously prompt it to dive deeper and deeper where you want it to go.

2. Structuring

AI is great at helping you to structure your ideas. You can feed it your messy notes or half-baked outline or even just a few scattered thoughts and ask it to arrange them into a logical flow. It’s likely not going to get it 100% right. But you can think of it like a scaffolding tool. It can help support your ideas so you can see the framework more clearly.

3. Gap filling

Let’s say you’re writing and you use a list of examples to show a point. But you can’t think of a third example. (And you like to have three because of the psychology of the Rule of Three!)

Asking an AI tool for suggestions of a third example is a fantastic use. It’s also very good at this. And if you don’t like what you get you can just ask for more suggestions.

This isn’t limited to small instances like this. You can use it for big picture issues as well. For example, if you’re outlining for a book and you can’t figure out a third point to support your main thesis, just ask ChatGPT for suggestions. It’s surprisingly good at big picture thinking (just be sure it’s not scraping the idea from another writer or thinker – I’ve seen this happen!).

4. Getting over writer’s block

Writer’s block or blank page syndrome happens to all of us. You’re ready to write but you just can’t get started. Instead, you find yourself just sit there staring at the page. My tip has always been to start in the middle, which helps you get over the worry about the first sentence or first paragraph. But now I often suggest that people turn to AI.

AI can help you get a first draft visualised quickly, so you’re not starting from a blank page. (But read on through the NOs before using this content as is!)

5. Editing assistant

AI can help you spot clunky sentences, grammatical errors or other proofreading errors. It can also suggest rewrites or help you simplify overly complex or jargony language.

Note – it is NOT good at working on long pieces of text. You’ll have to break it up into pieces and make sure you prompt well (i.e., do you use Oxford commas or not (I suggest not)?) I actually find that for books it’s just not helpful. You should use an editor then a proofreader who can see the big picture. But for a single section of a chapter, for example, it can be useful.

6. Simplifying the complex

I work on a lot of books with complex subject matter. Sometimes I just don’t understand what is being said. In those cases, one of my first port of calls is to go to ChatGPT, feed it the language and ask, What does this mean?

With the wealth of information in its LLM the AI tool almost always understands and can quickly summarise it for me. Of course, this is a place where you absolutely must double check as AI is not 100% accurate (as we’ll explore more below).

7. Example generation

When I work on books for authors, I often suggest including stories. I want them to include their stories of course, so I’ll use AI to generate an exemplar story so they can see what I’m looking for.

We won’t use this story in the end result, but it helps immensely with the editing process and saves me heaps of time explaining or drafting out an example myself.

8. Expanding on ideas

Sometimes it can be a challenge to capture the depth of our thinking. This is where AI is incredibly helpful. You can feed it a rough seed of an idea and ask it to expand. It might generate a few paragraphs, an analogy or even a list of examples you hadn’t considered. Again, this is now ready for you to refine – not to cut and paste (see the NOs for why that is).

9. Formulating introductions and conclusions

If you’re writing a nonfiction book you will very carefully craft the flow. That means you need strong introductions in each chapter that tie into the last chapter, tie into the main theme and introduce the main concepts of the current chapter. You’ll also need strong conclusions that tie back up to the introduction, sum up the content of the chapter and transition the reader into the next chapter. These can be time consuming to write.

AI is excellent at giving you drafts of this transition language. Not perfect, but excellent. And you then have a starting point to work from.

AI is a tool – YOU are the writer

There may be 1000 other ways that you can use AI in writing your own book – in fact I’m sure there are – and you’ll find the ones that work best for you. The above are just some ways that using AI has made my book writing more efficient.

But the unfortunate truth is that AI is not good at drafting long form content. The trick to remember is that it is a tool – a useful tool – but YOU are the writer. And you have to be the writer if you want to write a book worth reading. So on to when you can’t use AI in your writing

When the answer is NO

1. You cannot copy and paste AI content into your book

It’s literally mind numbing.

If you’ve ever tried to read a chapter of a book written solely by AI you’ll get this straight away. Despite how easy each sentence is to read – because AI is very clear and grammatically correct – the overall meaning is very hard to decipher. And that’s because it utilises patterns in its writing (that’s why AI checkers can recognise it) that literally cause our brains to fall asleep.

I don’t have any science to back this up (maybe I should commission a study – any volunteers?). All I have is my own experience reading thousands of books (personally) and hundreds of books (professionally) and writing and editing dozens of books as well. Whenever the content is written by AI I know within about half a page – not because I’m smart, but because I have no idea what I’m reading.

You know how when you reading in bed and you start to drift off and you suddenly you realise that you’ve read three pages but can’t remember anything? That’s what AI writing does in long form content.

The end result is that right now, and maybe it will change in the future with increased AI capabilities, YOU CANNOT USE AI WRITING IN YOUR BOOKS. It is boring. It is mind numbing. And nobody will understand what you’re saying.

Can you use a little bit? YES! Use the interesting phrase it comes up with. Use the cool metaphor. Take its grammatical suggestions. But by all that is holy, please do not copy and paste from AI into your book. I recently read a chapter that was about 62% AI written, and even that was too much.

Gary Provost, author of many books across a range of genres including four award-winning YA novels and a hugely respected writing instructor until his death, showed us better than I ever could with the below:

You lose the personalisation.

This article is pretty close to how I would talk to you. And any book that I would write would be the same. But if I gave it to AI what would it sound like? Let’s test it. I’m going to take the section that I wrote above and ask ChatGPT to rewrite it below (no additional prompts and no edits):

1. It’s literally mind-numbing

If you’ve ever tried to read a chapter written entirely by AI, you’ll know what I mean. Each sentence is crystal clear and grammatically perfect, but somehow the whole thing is harder to follow than a toddler explaining a dream. That’s because AI leans on repetitive patterns in its writing (which is also how AI detectors spot it). The effect? Our brains switch off.

I can’t hand you a neuroscience paper to prove this (maybe I should commission one — any takers?). What I can tell you, from reading thousands of books for pleasure and hundreds more for work, is that I can pick AI writing within three paragraphs. Not because I’m brilliant, but because I suddenly have no idea what I just read.

It’s the same feeling you get when you’re half-asleep, “reading” before bed, and realise you’ve skimmed three pages without taking in a word. That’s the AI effect in long-form writing.

So here’s the truth, at least for now: you cannot drop AI-written text straight into your book. It’s dull. It’s numbing. It’s unreadable. By all means, steal the odd phrase, a clever metaphor, or a neat grammatical suggestion. But if you paste AI text wholesale, your readers will notice — and not in a good way. I recently reviewed a manuscript that was about 60% AI-generated. Even that much was too much.

It’s exactly what I wrote in terms of substance. But it no longer sounds like me. And the more AI content you read, the more you see the same phrasings and patterns emerging, and the more recognisable it is as AI.

People are not going to engage with you and your message if they don’t feel you inside the writing. That’s just a fact. And it’s particularly true if you’re writing a business book that showcases you as a thought leader.

There are AI tools that are just for research (I’m looking at you Perplexity) but it still gets things wrong – and wrong a lot. You must do your own research. You must actually cite the article or study yourself  and confirm it’s from a trusted source. This includes research given to you in Google’s own AI results… these are often straight up wrong.

3. You cannot use it for all your ideas

If you don’t have any original thinking, don’t write a book. A book that’s scraped from what’s already on the internet is just a waste of time.

Of course, generative AI can recombine information into new ideas – that’s what makes it different from the AI we’ve had in the past. But it’s not really doing that when you work with it on a book. What it’s doing is tapping into all its learning to give you answers to queries. You would have to prompt really, really well to get it to come up with a new idea. And you can do this, in a brainstorming session. But you must be the one leading the charge. You are the thinker we care about.

4. You cannot give over your accountability

AI is great… but it’s not always right. The BBC completed research with four AI tools (ChatGPT, Copilot, Gemini and Perplexity) to see if they accurately represented BBC news stories when using them for source material. They found that nearly 20% of AI answers introduced factual errors in statements, numbers and dates. And 13% of the time, it just made things up.

That’s not good enough for your book. It’s not good enough for any of your writing. You are responsible for what goes out into the world. Under no circumstances can you say, ‘AI told me that was right!’ Anything AI tells you – from facts to research to stories – must be verified by you – the responsible human.

Here’s what the Harvard Data Science Review article, ‘Toward a Theory of AI Errors: Making Sense of Hallucinations, Catastrophic Failures and the Fallacy of Generative AI’, said:

‘What is becoming clear is that our new technologies are not only astonishing and incredibly powerful, but also fallible, inaccurate, and at times completely nonsensical. To make sense of this fallacy, the term ‘AI hallucination’ has become widespread in the tech industry, the media, and the public imaginary. Google, Microsoft, and OpenAI all publicly engaged with the problem. An internal Microsoft document stated that: “these systems are built to be persuasive, not truthful (…) This means that outputs can look very realistic but include statements that aren’t true.’

5. It cannot manage complex reasoning

Stanford researchers have found that complex reasoning remains a challenge for AI tools in 2025. They said, ‘AI models excel at tasks like International Mathematical Olympiad problems but still struggle with complex reasoning benchmarks… They often fail to reliably solve logic tasks even when probably correct solutions exist, limiting their effectiveness in high stakes settings where precision is critical.

Why does this matter for book writing? Because books aren’t strings of neat sentences. They’re built on arguments, insights, cause-and-effect thinking, frameworks and nuanced perspectives that require multiple steps of reasoning.

The problem is that AI is good at sounding like it’s reasoning. I can produce paragraphs with research, quotes and stories that look impressive at first glance. But as soon as you scratch the surface you’ll find the gaps, the holes and the embarrassing mistakes. You don’t want your readers to find those first.

Use AI in your book writing, but use it the right way

Here’s my honest truth from one writer to another – AI is an amazingly useful tool. I often describe it like having a junior writer at my shoulder. Someone I can bounce ideas off, or turn to and say, ‘what’s that word…’ or ask if there’s a metaphor about butterflies and bowling balls, and it will help me. But it’s actually a pretty crappy writer.

Use it prolifically as a helper, an assistant, a supporter, but not at all as an actual writer. As someone who’s read a ton of AI written content, I can promise you, no one (right now) wants to read that.