Author: Janelle Shane
Links: Bookshop (affiliate link) |Goodreads
Summary: This was a delightful, hilarious, clear introduction to AI that I’d recommend to anyone with any interest in the topic.
“‘You look like a thing and I love you’ is one of the best pickup lines ever… according to an artificial intelligence trained by [artificial intelligence researcher] Janelle Shane, creator of the popular blog AI Weirdness.” (source). Using other equally funny, accessible examples, Shane gives a fantastic intro to artificial intelligence. She identifies activities where we’re probably already interacting with some form of AI; explains how AI are trained; and shows where that can go wrong.
First, I’m going to make a quick terminology swap on you. The author starts out talking about AI, but that’s a pretty general term for “a system that seems smart.” She quickly moves into specifically talking about machine learning. This is typically the type of AI we mean when we’re talking about stuff that exists today. It’s basically a technique that lets a computer pick up patterns in tons of data, possibly patterns humans would miss. It can then use those patterns to make predictions and/or select a best next action. Machine learning is the term I’ll use in this review.
The chapter in this book that explained machine learning impressed me so much. The author does an incredible job giving clear, detailed descriptions of how different types of machine learning work. She does this at a conceptual level (no math involved!). I’m not a machine learning expert but know some basics and she was accurate to the best of my knowledge. The examples she used were hilarious and easy to follow. Comics interspersed with the text added to both the humor and the clarity. I think this chapter in particular would make an incredible addition to high school computer science classes. I’d also highly recommend it to anyone who wants a deeper, but non-technical understanding of AI/machine learning.
The rest of the book mostly focuses on the ways machine learning can go wrong. I think this would make a valuable intro for college students going into machine learning. It’s a great overview of mistakes to avoid and cases where machine learning isn’t the right answer. Some of my book club members found this book overly negative because of this focus. I do think the book should perhaps have been marketed more like Weapons of Math Destruction given this strong focus. I’m also not sure there was a single example of machine learning being used successfully in the real world. These examples exist; I just don’t think the author covered any of them. So the book did have a very particular focus.
The author’s examples of possible problems in machine learning are a tiny bit repetitive. A lot of the same principles pop up over and over. Still, the individual examples were so funny and enjoyable that I can’t say I ever got bored! I also find that repetition across books help me remember what I’ve learned so hopefully I’ll find the same thing happens within this book. That’s my only complaint and it’s a small one. I really can’t recommend this enough to anyone interested in this topic and think high schoolers and maybe even younger students could also get a lot out of this.
AI is something that I “get” but don’t really, if that makes sense. With all the new advances being publicized in the past few months I feel like I should learn more about it all.
That does make sense! And like you, it does seem like a timely read. I was glad my book club picked it to read 🙂