How I Wrote the Deep Learning AI Playbook

Photo by Alexa Mazzarello on Unsplash

Permit me the luxury of introducing a book I’ve been working on for the past year. For the readers of this blog, I’m sure you’ve seen this book mentioned frequently in my posts. However, I have yet to properly introduce the book to you and this is finally my opportunity to do so.

The book has its own website, which should be very easy to remember: “Deep Learning Playbook.com”. Every new content in today’s world has to compete for the eyeballs and the time of its customers. That’s where selling a book is a very difficult proposition, no matter how valuable the content may be. Who has the time to slog through 350 pages of text? Unfortunately, for complex and emerging subjects like Deep Learning, there aren’t any instant ways to download knowledge and wisdom into one’s own brains. If there was a way to make knowledge transfer even easier, I would very much like to know what that magic potion might be.

So ultimately, getting down to reading a book is a matter of priorities and urgency. The reason you might want to read my book over the more entertaining activity of binge-watching “Black Mirror” is that this playbook gives you a foundation of what is indeed real and what indeed is possible. For most people, technology is sufficiently advanced that it seems to be all magic. However, this is not the case and it’s important for everyone to be able distinguish magic from reality. Perhaps this is one reason I wrote this book, to discover for myself what’s underneath the magic of Deep Learning.

Ideas come from conversation

Writing a book takes considerable mental effort. This does not include all the other administrivia that is required for a book (i.e. cover design, layout, proof reading, copy editing etc.). The most difficult part of writing is a book is the generation of ideas. Fortunately, I have this blog and its many readers who have given me feedback, ideas and criticism that has helped me improve and fine tune new ideas. As a human being, it is very difficult to talk about something without receiving feedback. We are designed to appreciate conversation and by having an immediate conversation with your audience, you generate ideas. As in conversation, one idea becomes a link to another idea. When you write a blog, you get feedback and that in itself is a valuable motivator to get you to slog through the completion of a book.

Writing is knowledge discovery and knowledge discovery is iterative

A book is always never really finished. You will always find new and more compelling concepts and fresh ways to present content. So as it is best practice in agile environments, it’s always best to confine scope by time-boxing the process. Also, it’s always better than to be late. Knowledge discovery will always be an iterative process and book writing is essentially a knowledge discovery process. It’s very rare that when you start writing a book, you already know all of what needs to be written. Usually you discover what you don’t know when you begin to write.

An entrepreneurial mindset that is always seeking for opportunities (or ideas) is key

Forcing oneself to write is like photography. When you are into photography, you’ve consciously decided to capture the best angles of an event or place you find yourself in. This makes you more consciously aware of the environment than a casual visitor. You are aware of the lighting, shadows, interaction of people, shooting angles etc. It is the same with writing. You put yourself in a perspective that many may not be aware of. That is why, photographers can learn from other photographers and writers can learn from other writers. Thus, I am grateful for the perspectives other writers have lent me.

Don’t just write one book, write two or more!

In the Deep Learning Playbook, I introduce this idea of Jobs To Be Done (JTBD). It’s a way of understanding the kind of product you need to build to satisfy a customer. In essence, you should create products that make it easier for people to accomplish their jobs. The complexity however comes from the different motivations or jobs of different people. Thus, it’s inefficient to create a product that is tailored for too many. It dilutes the message and you have a product that’s just too generic nobody understands how it would be helpful.

The Playbook’s audience are people in business who need to know what to do to take advantage of this latest A.I. technology known as Deep Learning. So, if that’s what you’re doing in your current job, then it makes sense to get this book. However, most people only want to know about Deep Learning from a 30,000 foot level. How do you address the people who are just curious and want to just expand their knowledge? Furthermore, what if one of your goals is to introduce to as many people an idea that sounds like an oxymoron (i.e. Artificial Intuition)? Well that’s why I came up with another book that’s affordable to the widest audience possible:

Explore Deep Learning: Artificial Intuition: The Improbable Deep Learning Revolution

Compared to the Playbook, I truly enjoyed writing this one. However, like many things in life, the stuff you enjoy doing isn’t the stuff that people will pay for. People tend to pay for the more expensive Deep Learning Playbook. It is simple economics: if one can monetize what they are buying then its price is less of an issue. That’s why financial products like stocks and bitcoin are easy sells. Everyone wants to buy a ‘money printing device’.

You are lucky if people have time to write a testimonial

A book is always best sold with testimonials. I am fortunate to have found people who have taken the time to go through my book and lend their impressions. Here are a few of the notable ones:

I love it. Chapters 3, 4, 8 and 9. and your approach thru the lens of intuition!
— John Seely Brown, Author and co-founder of the Institute for Research on Learning
Then it continues to the principles and best practices, followed by cutting-edge research. Many of the ideas actually lead towards what we call AGI (general AI) The book takes a deep dive into “meta learning (learning to learn)”, which I believe is the most efficient way to automate engineering of a thinking machine: bootstrapping itself and recursively self-improving its adaptability.”
- Marek Rosa, CTO GoodAI
His playbook has given me great inspiration on the latest topics and viewpoints for my deep learning lectures. Our students and alumni consider this a helpful reference and strategy guide as they find new uses for deep learning and AI in industry.”
- Ellick Chan
Adjunct Lecturer MSiA 432: Deep Learning
Northwestern University
His is a refreshingly different approach to AI It is easy to read and at the same time covers a lot of complexity and detail.
- Ajit Jaokar, Director of AI / Deep Learning Lab for Future Cities, University of Madrid

Writing books are for the passionate and not those seeking financial rewards

Finally, a word about the economics of publishing a book. Most people who have never written a book are unaware of the brutal economics. I have chosen to self-publish. Gumroad is a god-send for writers in that at least 95% of the purchase price goes to the original author. However, when a writer goes with a publisher, he would find himself lucky to even receive 12% of the net revenue. A printed book has its own cost, a standard “6x9” paperback costs around $5 to produce. A hardbound version, costs a lot more (i.e. $12). After that, the distributors are also going to get their cut. So depending on which distributor you use, they can get at least a 40%-65% cut from the sales price. So, just to put this in perspective, a writer who sells a book via a publisher for $50 should expect to receive a paltry $3.24 ( (50–5)*.6*.12) in revenue. Try feeding your family on those wages! Very little goes to the author will all these middle men taking their cut. So don’t go into writing a book for the money. Write because you are passionate about the subject. And if you want to be fairly compensated, get rid of the middle men. That’s what I did and I enjoyed the entire process!

Exploit Deep Learning: The Deep Learning AI Playbook