Publishing a book remains one of the best ways for an executive to stand out. As the 2024 Business Book ROI study found, when done well, a book can sharpen a leader’s market position, boost their brand, help achieve business goals and ultimately turn a profit. And leaders who offer insightful, authentic writing amid the onslaught of AI-generated content will form deeper connections with readers and potential customers.

Authors can leverage AI as a tool to reduce the time and cost of writing a book. It can organize transcripts, summarize research, structure chapters and produce promotional copy in minutes. But while AI is opening more doors for authorship, a completely AI-driven process can actually decrease your potential return.

Having worked on around about 10 books using AI in some way, there are some clear use cases—and some that will create massive headaches:

• Brainstorming: AI can be helpful, but don’t rely on it too much.
• Outlining: A strong use-case for AI.
• Research: AI can be a timesaver and a lifesaver! Trust, but verify.
• Writing: Tread lightly…
• Editing: Tread very lightly…
• Miscellaneous tasks and promotion: Depends on what you want to do, but AI can be a huge help.

Start with these two general rules of thumb:

• AI can do less, not more. If you expect AI to draft a 70,000-word manuscript from a one-page outline, sorry—that’s not happening. AI struggles to expand text in a coherent way, but it is very good at taking a lot of text and producing shorter prose. (Why would you want to go smaller when writing a book? More on that below.)

• Use AI as an over-eager intern. Many people have observed that AI is excited to serve you but does not have the capabilities or knowledge you might expect of a person with its level of confidence. Do not give AI any task you wouldn’t give a 22-year-old college grad.

1. Brainstorming

Lots of people use AI in their ideation processes. A sparring partner, trained on the entire knowledge of the internet, could only help sharpen your ideas—right?

Unfortunately, after seeing several authors brainstorm with AI, and using it myself on occasion, I think “ideation” could be the worst use case for AI. Its feedback is not situated within the full context of your thought leadership. Telling it to be “more critical” can lead to it picking ideas at random and tearing them apart for no reason. It does not understand the logical structure of what you want to communicate, so it cannot help strengthen your ideas in a coherent way.

There are some exceptions to this: If you are brainstorming for examples, or specific research, AI can be quite helpful with the right prompts, much like an intern directed to explore the internet for a day, albeit with results in just a few seconds.

2. Outlining

AI is fantastic at outlining—if you give it the right inputs. When I work with someone on a book as a collaborator, I start with in-depth interviews. This produces a lot of ideas, recorded in transcripts. It takes considerable time to reduce all of this down into a clear outline for a book or a chapter. AI can do it much faster, often in seconds. This is when “more to less” is a huge help.

You do not even need to be interviewed about your book to benefit from AI’s analytical skills (although it helps to have a human sparring partner, see above). I worked with someone who had hours of recordings of him talking to himself about the book that a research assistant spent days going through, but it could have been synthesized by AI in no more than a few minutes.

3. Research

Research is a common use case for AI and I can confirm AI can be a huge timesaver for business book research. It’s an intern-level research assistant at your fingertips. Make sure you are verifying the information (like you would do with any research assistant), but in my experience, there has been less AI “hallucination” in the last few months than two years ago.

4. Writing

Using AI to write your book is a sticky ethical problem. If AI “writes” the book, can you claim ownership of it? (As of right now, probably not.) But that’s less important than AI’s current inability to write compelling prose beyond a few sentences. If you want a book, AI can write it. If you want a coherent book, AI cannot.

There are some caveats to this: AI can generate portions of text that might be useful, with the right prompting and context. If you upload a series of case studies from your company, for example, it can synthesize those into a paragraph or two for an example in a chapter. I worked with someone on a highly technical book and we uploaded probably five times the amount of text required for a book-length manuscript (including manuscript drafts we had previously written and a variety of articles), and AI delivered a decent draft.

An AI draft can be a starting point (or what Anne Lamott calls a “shitty first draft.”) Getting to that point can be the hardest hurdle to overcome. But it will likely then need to be revised with more compelling and logically sound prose. Relying solely on AI for writing robs you of the thinking that goes into writing, which will help you make a better book, which will help you connect with readers and create more opportunities.

5. Editing

I’ve assumed in the past that because spelling and grammar is dictated by clear rules, AI could be a competent book editor. But unfortunately, the massive amounts of text, and the complexity of sentences, seem to confuse the technology. It makes up edits, ignores glaring problems or stops halfway through. Maybe AI can improve on this‚ but it’s not there yet.

6. Miscellaneous tasks and promotion

AI can help with many other tasks that go into book writing: formatting references; creating indexes; generating graphics, especially diagrams or charts. These tasks can take up a lot of time (and can be the rote, boring tasks an intern could take on).

AI can also create the supplemental content required for promotion: social media posts, newsletters, collateral, launch event content and more. For example, if I upload a chapter into AI, it can create a reasonable draft of a teaser newsletter to promote the book.

Business leaders should not be afraid to use AI in their book writing process. Used well, it can save time, reduce cost and improve organization. But it should not be allowed to replace the thing readers actually want from a leader: their unique thought leadership and judgment.