AI as Assistant, Not Author: A Tequila & Tech Roundtable on Writing, Music & the Human Heart

This Tequila and Tech Talk was a recurring high-level roundtable that Richard Lowe (The Writing King) co-hosted with Natan Verkhovsky, a gathering of cybersecurity and tech executives, writers, and creators trading ideas over a drink. This March 2025 session took on AI directly: where it helps, where it fails, and why every practitioner in the room, across writing, music, and engineering, drew the same line between assistance and authorship. The series is no longer running, but the conversation holds up as a clear-eyed practitioner’s view of AI’s real role in creative and technical work.

Natan: Let me introduce Richard. If you know him on a technician level, that’s a cool Richard in another life. In the most evolved version, Richard has become a ghostwriter with over 113 books written. So if you ever wanted to be a TED talk speaker, apply for venture capital and succeed, get on the circuit as a paid speaker, or parlay your expertise in AI, cybersecurity, cloud, or blockchain to go from VP to C-suite, these are the past benefits Richard comes with. Richard, if I left any meat off the bones, fill it in.

Richard Lowe is The Writing King, a professional ghostwriter and author who has written more than 113 books under his own name and ghostwritten over 54 more for clients.

Before writing full time he spent 33 years in enterprise IT, including two decades as Director of Computer Operations at Trader Joe’s.

DISCUSS YOUR BOOK

Full transcript of the conversation follows.

From Trader Joe’s Cybersecurity to Ghostwriting

Natan: Let me introduce Richard. If you know him on a technician level, that’s a cool Richard in another life. In the most evolved version, Richard has become a ghostwriter with over 113 books written. So if you ever wanted to be a TED talk speaker, apply for venture capital and succeed, get on the circuit as a paid speaker, or parlay your expertise in AI, cybersecurity, cloud, or blockchain to go from VP to C-suite, these are the past benefits Richard comes with. Richard, if I left any meat off the bones, fill it in.

Richard: I was in tech for 33 years, been a ghostwriter for 13, so I’ve been doing things for a long time. I was the Director of Computer Operations at Trader Joe’s for 20 years, which basically means I did all the stuff nobody else wanted to do: cybersecurity, disaster recovery, you name it. I was also a photographer, about 1,200 shows and dance events and 300+ Renaissance faires. Then I moved from California to Florida, because California got a little weird for me and Florida is quieter. The only things in the backyard are alligators, and the bobcat in the tree, so don’t leave your cats out at night.

Natan: Did we want to put an AI slant on tonight?

Richard: Sure, why not.

AI as a Digital Assistant, Not a Writer

Richard: I’ve written several books on digital transformation and AI. The thing AI is good for is being a digital assistant. You don’t use it to replace people, you use it to enhance them so they can get more done. As a writer, I use AI as a digital assistant. If I’m writing a novel for somebody, I might put in a chapter and ask: are the characters consistent, is this character acting out of character? On a nonfiction book: am I being redundant anywhere? I interrogate it and it sends back answers. I don’t write with it, because that’s not what it’s for. It’s a terrible writer. It’s flat. It’s boring.

Ken: There’s no such thing as AI, because natural intelligence hasn’t been discovered yet, so there can’t be artificial intelligence.

Richard: Thank you, Ken. If you use it in that context, giving people more capability, you can’t go wrong. If you’ve got an accounting person, you give them a tool to do their accounting job better.

The Uncanny Valley in Writing

Richard: If any of you know robotics, there’s a term called the uncanny valley: you look at a robot that’s supposed to be human, but something’s off. It’s especially prevalent in CGI. Watch the new Lion King remake and there’s something wrong with it, you can’t quite figure out what. The shadows might be off, or the hair flows the wrong way in the wind. Your conscious mind doesn’t catch it, but your unconscious mind is totally onto it. Writing is the same way. Your unconscious mind reads AI text and goes, this just doesn’t make sense, there’s something weird about it. That’s what makes your audience put the book down.

Just a few days ago I gave somebody an hour of coaching, he handed me his book, and I opened it and said, you wrote this with AI, just about that quick. He said, yeah, how’d you know? Uncanny valley. He rewrote parts of it and published it.

The Same Lesson in Music Production

Scott Weber: I like what you said about AI, because for ghostwriting there are a lot of similarities with music production. As a rule, I only use AI as a way to enhance the workflow, never as a replacement for vocals, guitar solos, arrangements, anything unique to an artist. Those are all me. I currently use it for mastering, the Isotope Suite has an AI assistant that analyzes your sound and comes up with a good EQ balance. It’s a good starting point, but nowhere near a replacement.

Richard: You take out the boring stuff. Chomping through a book to find the characterization that’s off, AI can do that instantly, where it would take you days. Workflow is an excellent use for AI in any industry.

Scott: And like books, you can tell AI-produced music. People in professional studios can tell the difference, and they’ll be turned off.

Richard: The authors and musicians using AI instead of their own creativity aren’t going to make it. They might initially because it’s new, but there’s so much AI garbage on the internet now that people are starting to ignore it.

AI for Citations, Consistency, and Catching Gaps

Ken: I do a lot of work with this in both professional and academic settings, and you’re absolutely right. It’s great as an assistant to tell you what’s missing. Are there inconsistencies compared to other sources? Any gaps in what I’m presenting? Don’t write it for me, just tell me what’s wrong. Am I being stupid in some places? It’s very good for pointing things out. But a human being has to go in there and put the heart and soul into it.

Richard: One thing it’s very good at is citations. I ask, is there anything in here I need to cite? It tells me, cite this, cite this. Sometimes it’s wrong, but it gives me a start on all the citations the client forgot to mention they took from somewhere else, so I can make sure it’s legal and ethical.

Spotting AI Writing: The Em-Dash Tell

Tim Maile: I keep telling people the same thing about using it as a tool. They say, I can do all your marketing with an AI thing, then they come to me because it’s gone wrong and got them in Facebook jail or LinkedIn jail. I can spot AI-generated content already, because it sticks to the same phrases.

John Epstein: There’s this whole conversation about the prevalence of em dashes among the language models.

Richard: It seems to love them.

John: I like a good em dash. I was like, come on.

Richard: The AI-detection tools check for that. It’s definitely an AI tell, because hardly anyone uses em dashes the way the models do. It’s generally easy to tell.

John: The email outreach it generates, I just had to rewrite it. It was too hokey. I can tell when someone uses AI to write their spammy LinkedIn outreach to me.

From Dungeon and Zork to Activision

Tim Maile: I’ve been a PC gamer for 30 plus years. I turned a hobby into a business.

Richard: My first computer game was called Dungeon, the predecessor to Zork, written in Fortran, and it was the most fun game I’ve ever played even though it had no graphics at all. All text driven: go south, go north, get weapon, kill troll. It’s a puzzle game, you had to get 500-some points, and I could never get the last point even though I had the source code. I started on the TOPS-10, a 1970s machine. The only game before that was Adventure, also called Colossal Cave, that Dungeon was based on. And the company that made Zork eventually morphed into Activision, which you’ve probably heard of.

The Afghan Politician and the Backstory That Never Made the Book

Natan: Were there areas that surprised you, working with clients from technical backgrounds?

Richard: A lot of the stories surprised me, how people got into their industry. The backstory is often not in the final product. The first book I ghostwrote was for an Afghan politician who lived there before the Taliban existed. He was responsible for all the roads in Afghanistan, and they were going to assassinate him when they took over. He had an escape story that was just, wow. The interesting thing was he didn’t speak English, so I interviewed through his wife, who barely did, and he wanted to go back to Afghanistan, so I had to do all the interviews in one day, 16 hours straight. It surprised me that I could do that. It was way out of my box, and I had to learn Afghani and Muslim culture I didn’t know.

Taking the Heart and Giving It Back

Richard: The point is I put that backstory into the work. If it’s a book on technology, AI can write a dry, boring paper all you want, that’s fine for that. But if you want something that stands out, you want your heart in it. I take your heart and your soul and put it into the book, then give the heart back. AI can’t do that, probably never will until it gains sentience, at which point the game changes. One of my clients got a TED talk from it. Another got $30 million in venture capital. Most of them get paid speaking engagements. All because they wrote a book and I put the heart into it.

AI in Cybersecurity: False Positives and Attack Surfaces

John: In our cybersecurity work, we use the language models to generate response playbooks, but the main use is false-positive reduction, which is a totally different thing from writing.

Ken: From a system-security point of view, if you talk about the system and its requirements and ask about attack surfaces or potential vulnerabilities, it may come up with some interesting ideas. I recommend trying that sometime.

Richard: One reason it hallucinates is its training data. Its base includes a lot of internet text that isn’t famous for being accurate, so it inherits that style and the false information in it.

Closing Reflections

John: This was an enjoyable session, learning from very intelligent, experienced executives sharing thoughts about the future of these technologies.

Ken: It’s always valuable to hear tips and techniques. Very good tips for using large language models like ChatGPT to date.

Royce Blake: I’m amazed at the level of intelligence in this gathering. I’ve written for 17 different industries, all unique, and it always amazes me, right when I think I’ve heard it all, somebody comes out of the woodwork.

Richard: I’m running a Leaders and Their Stories podcast, daily, and I’m actively looking for people who want to speak for 20 minutes about their story. I used to hate video and be super shy, and Natan has dragged me into the 21st century.

Find Richard Lowe at TheWritingKing.com.

Quotable moments

Frequently asked questions

Should you use AI to write a book?

No, but you should use it to assist. The panel’s consensus, from a ghostwriter, a music producer, and an engineering professor alike, is that AI is a strong assistant and a weak creator. It’s excellent for checking consistency, flagging redundancy, finding gaps, and suggesting citations, but its actual prose is flat and boring, and it can’t supply the heart, voice, or original creativity that makes a book worth reading.

What is the uncanny valley in writing?

It’s the borrowed robotics term for when something is supposed to read as human but feels subtly off. With AI text, the reader’s conscious mind can’t always name the problem, but the unconscious mind catches it, and that vague wrongness makes audiences put the book down. Richard can spot AI-written books in seconds for exactly this reason, and he warns authors that the effect quietly costs them readers.

How can you tell if writing was generated by AI?

Several tells: it sticks to the same recycled phrases, it reads as hokey in outreach and email, and, as the panel noted, it overuses the em dash to a degree most human writers don’t. AI-detection tools flag these patterns. The deeper tell is the uncanny-valley feeling, the sense that something is wrong even when you can’t point to the exact sentence.

Where is AI genuinely useful?

In the unglamorous, repetitive work. For writers: consistency checks, redundancy detection, citation suggestions. For musicians: mastering assistance and EQ starting points. For cybersecurity teams: false-positive reduction and surfacing potential attack surfaces and vulnerabilities. In every case it accelerates the work a human still owns, rather than replacing the human.