Too many technologist, in every generation of technology, state that management need to think more like programmers. That’s not the case. Rather, the technology professionals need to learn to speak to management. “The Business Case for AI,” by Kavita Ganesan, PhD, is a good overview for managers wishing to understand and control the complexities of implementing artificial intelligence (AI) systems in businesses.
I’m always skeptical of self-published books. Usually that means the books just aren’t that good. However, sometimes, especially in non-fiction, it means that publishers are clueless about the subject and hesitant to work with people who aren’t “names.” This book is an example of the second option, and it will give management an introduction to the concepts surrounding AI and how to address implementation in a way that will increase the odds of success for AI initiatives.
The indication that the author mostly lives in the real world comes quickly. The first chapter is a good, introduction to what matters for business about AI. Forget the technical focus, it’s about solving problems in an efficient and cost-effective way.
Chapter 2, “What is AI?” isn’t bad either, though I disagree with the idea that machine learning (ML) is part of AI. Business Intelligence (BI) has advanced, along with computing performance, that standard analytics provide insight that can be termed ML, so ML and AI overlap. That, however, is a religious argument and what Ganeson has to say about AI is at a good level for management understanding.
The weakest chapter in the introduction is the fourth, where the science fiction addict in me had to sigh at “Movies such as ‘I Robot’”. Ummm, check your library.
That chapter’s list of myths is also a bit problematic. The first, about job loss, is the one area where it shows the part of the real world in which the author exists isn’t one most people are in. The AI revolution is very different than the industrial revolution and earlier technology revolutions. She talks about artificial general intelligence (AGI) and says that since it’s still far away that means a lot of jobs won’t be lost. We don’t need AGI to replace jobs.
The next couple of chapters are good for setting up examples of business processes that could be impacted by AI. I do have an issue with which companies she decides to name and which remain anonymous, as that seems to imply protecting customers. The best part was a good discussion of IT & manufacturing operations, but that could have been improved by discussing infrastructure operations such as pipelines and the electrical grid.
Part 3 (chapters 7-9) is very good but, again, has a few things to keep in mind. On page 117, six phases of the development lifecycle are defined. I agree with them, but want to point out that data acquisition and model development, phases 2 & 3, can be done somewhat in parallel. The things you learn from each can impact the other. The other nit is that the author seems to use warehouse improperly. Data warehouses have a specific, more narrow purpose. When she uses the term, think data lake. The importance of logging, transactions and more, is often ignored, and the end of this section of the book has a good explanation of its importance.
The fourth part of the book is a set of chapters that drills down into the “finding AI projects” portion of the analysis process, and is well laid out.
The final section has two chapters. The first is about “build v buy.” It is no surprise that a consultant leans towards build, that’s her livelihood. What managers need to understand is that businesses aren’t as unique as they wish to think. There are unique things, but the vast majority of business is like other businesses. AI is a new technology and there aren’t enough easy to use tools for a buy decision in many areas, but that will change over time. Managers need have a flexible understanding of the equation and balance that in the real-world.
The final chapter is, as expected, a good summation and a return to focusing on business results. It continues the author’s use of good, simple, graphics to show the points of her arguments. Regardless of the issues I’ve mentioned above, the book does a great job of laying out the challenges of artificial intelligence from a business perspective. The book doesn’t delve deeply into algorithms or other details that don’t matter to management, while it does provide a framework to look at AI projects through a business lens that integrates the technology into organizations in a way that doesn’t leave everything to technologist. “The Business Case for AI” is a good introduction for IT and line managers to think about how to integrate artificial intelligence into their organizations.