“The New Fire” – The Book Doesn’t Live Up To The Promise Of Its Subtitle

After finishing “The New Fire: War, Peace, and Democracy in the Age of AI,” by Ben Buchanan and Andrew Imbrie, I am disappointed. There’s a strong promise in the title, but the book falls far short of that promise. There are a couple of interesting points in the book, but it’s not worth the read.

Those who have followed this column have seen book reviews over the years. I’ve been send a number of books on artificial intelligence (AI) and machine learning (ML) What you haven’t seen is a negative review. I’ve read some terrible books, and I’ve proceeded by ignoring them. This is the first time I’ve decided to write a negative review. It’s not that this is terrible, I’ve read much worse, but I’ve decided it is time to point out when there are ones that should be avoided. This is one.

The subtitle suggests a focus on international relations, and there is some coverage in the book. The key problem is that it takes a while. The first four chapters are in a section that is supposed to provide background. The section doesn’t do a good job. The first chapter is the one that drove my recent rant about deep learning and magic. The second discusses algorithms with a few, very standard, examples without describing what algorithms are or why they matter. The third is focused on hardware. It mentions the switch to GPUs, but doesn’t state why that happened. It wouldn’t have taken long to describe the similarity between arrays for screen processing and arrays for neural networks, but that didn’t occur. The fourth chapter was about the need for opacity. It’s important, but the chapter stays at a very vapid level of text, with buzzwords but not a lot to consider.

The next four chapters are supposed to be the focus. Chapter five is “inventing,” but not until one hundred pages into a slightly more than three hundred page book, do the authors finally get to a mention of artificial intelligence in the context of the military. However, almost the entire focus is on Google and the Maven project. It’s a case study of one instance without a broader stroke picture. The one key takeaway is how neither executives nor employees come off looking all that good. Executives seem to have no interest in the social good, while supposedly liberal employees seem to be saying corporations should be more powerful than nations and impose upon the rather than citizens working within their nations to discuss issues.

Chapter six is ​​“Killer,” and finally discusses weaponry. The first part of the chapter is again focused on Maven and superficiality, but the second half is the best pair of the book. That half contains a good discussion about autonomous systems and their likely use by democracies and autocratic regimes. There’s also a good analogy comparing how the US government viewed open submarine warfare before and after Pearl Harbor, and how governments might view autonomous systems based on conflict situations.

Chapter seven has a standard discussion of using AI/ML for hacking; but the best part is the latter half, with a nice discussion of the potential of hacking AI/ML systems.

That chapter and the next are, I’m guessing, supposed to be about the “democracy” portion of the subtitle, but they are written too broadly. One of the more intriguing segments of chapter eight is part of their continued focus almost exclusively on Google. The book was published this year, so it was, most likely, written in 2021. On page 192, text reads “Since 2016, the company has on more than one hundred occasions blocked users and close down groups.” That’s written as if it’s a good thing, but in five years that global organization has dealt with “more than one hundred”? Not two hundred, not a thousand, not a significantly large number given the size of the organization’s global presence. More than one hundred. That’s a drop in the ocean.

The last two chapters are “Fear” and “Hope.” I fear there’s not much to hope for there, and I skimmed those chapters. You will note that the one part of the subtitle I haven’t mentioned is “peace.” There is not enough coverage of how AI can be used for nations to understand each other and to negotiate, only some sections that discuss intelligence agencies and finding out things other nations are doing. Translating that into actionable diplomacy to help keep the peace is important but missing.

As you can see, there are things to like in the book, but those might have been better covered in articles or white papers. If you are interested in this book, I strongly suggest checking with a library, as my opinion is that this isn’t worth you spending anything other than a small fraction of your tax dollars on this book.

Leave a Comment