“Generative AI in Practice” is a good book. Ignore the buzzwords.

I hadn't been sent a book to review in a while, so I decided to take the plunge and give this book a try. To my surprise, I can wholeheartedly recommend it. “Generative AI in Practice” by Bernard Marr (Wiley, ISBN: 978-1-394-25424-8) is a good book if you ignore the first word.

Let's start with the speech. I've been avoiding anything with the word “generative” in it since it became a buzzword in the current hype cycle. This has been presented as a revolution in artificial intelligence (AI) because so many people like the idea of ​​revolution. In reality, evolution happens much more frequently. ChatGPT and other large-scale language models have shown that much more complex deep learning systems, powered by the cloud, have advanced the technology. And that's it. But if you put a cool adjective in front of AI, you can ask for more funding.

What's great about this book is that it provides a very clear overview of the current state of AI in business, and if you can ignore the hype, it's perfect for that purpose.

As is common in these books, skip the first few chapters. The author does not understand the history of AI and tries to argue, using chess as an example, that “traditional” AI followed existing rules. But the Go system shows that previous systems had already come up with tactics that people hadn't thought of but that worked. And the author tries to suggest that deep learning, i.e. neural networks, magically appeared with ChatGPT and generative add-ons. This is nothing new. As I mentioned in the last decade, what has changed is the power of servers in data centers, i.e. the cloud. We can now run more complex models much faster, so of course things are progressing.

I also laughed at the assertion that with modern systems, “you don't need to be a data scientist to explore data.” Really? Not only did AI do that earlier, but Business Intelligence (BI) systems have been doing it for a long time before that.

Okay, without further ado, let's talk about why this book is so enjoyable.

There is some standard material in Chapter 4, but one very important element is one he has covered elsewhere for several years: The plan is for companies to use AI to take over the simple, rudimentary tasks that many industries (such as insurance) have. The argument from management is that this will free up time for humans to handle the more complex tasks. But how will new employees learn if they don't start with simple tasks and then work their way up to the complex ones? It's clear that the CxO suite and shareholders want to automate everything, but how will the transition be handled so that employees can be trained while they need to?

Chapter 5 brings up another important issue. Marr is the first business management author to directly address the huge job disruption that will occur. My only complaint is that he doesn't go deep enough. Again, AI is limited to a scope not far from what we currently understand. Indeed, even researchers will be replaced, as Chapter 13 suggests is already happening with molecular research on new compounds.

All the chapters in Part 2 are great for a quick skim. Each chapter takes a business sector and explains how AI is evolving them. Again, ignore his repeated references to the generative element; he's a self-described “futurist” and “influencer,” so he has to get all hyped up. Still, he gives a good overview of what's already happening with AI in each sector, and what could happen.

Any manager will benefit from understanding the broader impact of AI on business, not just from specific examples from their own industry. Systemic forces will change how our economies work, from local to global. The book doesn't address regulations or the major societal issues that will need to be addressed as more jobs are automated. But it's a valuable, quick read for a good overview of the state of the art of AI in business.