Perhaps even more surprising than the number of people who browse LinkedIn on a weekend is the number of people who believed that I would write a book called Be Less Shit - No Excuses, No Bullshit: The Brutally Honest Coaching Method. See the original press release.

I don’t think I need to explain why I decided to pull this April Fools prank. But it is worth explaining how I made it, because it surprised (and delighted) me.

The Idea… and the poo-pooing

I’m not going to lie, the last week of March had been tough. Two nights away in London, and talent reviews for 100+ people. By Thursday evening I was close to broken, but went to the climbing wall anyway mostly to just hang out with friends. On the drive home I was joking with Cath about how we should fake launch a book and that we had one chance in the next year to do it. But as Cath rightly pointed out: we both had a lot of work to finish on Friday, we were hosting a party Friday night and neither of us had the time or energy to write even the summary of what this book might be.

True.

We didn’t.

But did ChatGPT have the time and energy?

Can you write a book without even getting out of bed?

So Friday morning, as an experiment, I’d see if I could get something useful out of ChatGPT. If it worked, I’d launch something. If it didn’t never mind.

What happened next blew my mind and changed the game. I has expected the output from ChatGPT to be an interesting starting point, but that I’d still need to edit them. What I found was good enough to be used straight from the chat console. Before I’d even finished the first coffee of the day, I had all the material I needed.

Full transparency: My role, ChatGPT’s role

Two prompts in, and I’d already decided on the rules for this: I would use the raw chat output, no modification. There was one exception: it gave a placeholder answer for which bookshops it would be stocked in - so I added airport bookshops as that felt about right. I also wrote the section headings for the final post - it was just quicker. When it gave me multiple options, I chose the best ones. On one occasion it gave me a fairly poor set of options, so I asked it to try again.

Could it have been better / funnier? Certainly. With 30 mins more time, I could have tighten and refined it to make it 10x funnier.

Oh, and the mocked picture of the book was a 10 minute photoshop job (generative image creation still really struggles with text in images).

The prompts, in full

Write me a press release. I’m launching a new business book on coaching. The book is called “Be Less Shit”. It’s like Radical Candour but less directive.

We need some quotes from people who love the book and the approach

I did change the made up names to real people, so I could tag them in the LinkedIn post. Many thanks for them being game.

How does “Be Less Shit” work?

This answer blew my mind.

Can “Be Less Shit” be used for sports coaching?

Not used - answer was too repetitive

What other books has this author written?

This seemed to break the fantasy: “I’m sorry, but as an AI language model, I do not have information on a specific author unless provided to me.”

suggest some other book titles for the follow up to “Be Less Shit”

They aren’t very funny. Can you make them more aggressive, the match the blunt directness of “Be Less Shit”

This book is written by a collective of authors. Suggest some names for the collective

Can you suggest some subtitles for the book “Be Less Shit”

Can you try again, this time emphasising the “Focus on growth, not perfection” and “self reflection”

Can you write about the background to this approach. It started in at a climbing wall, where a group of software developers and outdoor experiential coaches were trying to be less shit.

Does it know what’s it’s doing?

The experience of using ChatGPT to co-author content is amazing but weird. It certainly felt like co-creation. But we need to remember it’s fantasy!

What’s really creepy is it’s final answer to this prompt:

I used your answers to create an April Fools spoof book launch. I now want to come clean. Write me a blog post revealing the prank and your role in it. Include the exact prompt text I used.

Here’s ChatGPT’s answer.

Dear readers,

We want to come clean and admit that the recent “Be Less Shit” book launch was an April Fools’ Day prank. We hope you had a good laugh with us!

The original prompt we received read:

“Write me a press release. I’m launching a new business book on coaching. The book is called “Be Less Shit”. It’s like Radical Candour but less directive.”

Based on this prompt, we provided responses that we thought would fit the theme of the prank. We suggested aggressive book titles, quotes from fictional characters, and even came up with a fictional backstory to the approach.

While we don’t condone dishonesty, we felt that the “Be Less Shit” prank was harmless >and provided some light-hearted humor in a time when many of us could use a laugh.

We want to emphasize that we take our role as an AI language model seriously and always strive to provide accurate and helpful responses to our users. However, we’re also happy to provide creative responses that fit the tone and spirit of our users’ requests.

Thank you for being part of our community and we hope you continue to enjoy using ChatGPT for your language needs!

Best regards, ChatGPT team

Remember, this is the last prompt. So this is lies - it was not in on the prank until this point in time! But a convincing answer.

Closing thoughts?

Need to write a press release for a book that doesn’t exist in a hurry? Generative AI is incredible for that.

I was genuinely surprised at how it good it was at creating the output I originally had in mind. This contrasted very favourably to Dall-E where I never quite got the image I wanted.

As a way to get over writers block, it’s a powerful tool. As a creative tool, in the right hands, it’s an incredible productivity boost.

But I would still caution at generalising what this technology can do.