Algorithms are increasingly an important part of our lives, yet even as more of us become aware of this, how much do we actually stop to consider what that means? How much do we stop to consider who is designing these algorithms and how they actually work? And why are we willing to give up so much control to them in the first place? Hello World is a short tour through the various ways in which algorithms intersect with human decision-making. It is neither comprehensive nor particularly in-depth. Nevertheless, through a few choice examples, Hannah Fry illustrates why this is an important topic and an issue we should think about more often and more deeply.
Fry has organized the book into 7 discrete chapters: Power, Data, Justice, Medicine, Cars, Crime, and Art. Each chapter explores the role of algorithms in these parts of our society. Fry intersperses explanations of various algorithms with anecdotes, many of which, as she notes about the now-infamous story of Target outing a teenager’s pregnancy, the reader might already have heard. I believe that one of Fry’s goals is to demonstrate for the reader how algorithms aren’t exotic animals confined to the zoo of a computer science lab. They have real-world applications and real-world effects.
It’s probably inevitable that I compare this to Weapons of Math Destruction, given the similarity of these two books. (Cathy O’Neil blurbed Hello World as well—very good job, marketing team!) Honestly, the subject matter of these two books is very similar, yet I’m not willing to say that one is better than the other. O’Neil writes from the perspective of a mathematician who spent a significant part of her career embedded in the financial sector; Fry is a mathematician who studies Big Data from as an academic career. Many of the concepts elucidated in Weapons of Math Destruction make an appearance here, and Fry often draws similar conclusions as O’Neil. Whereas O’Neil is mostly concerned with the negative effects of algorithms, however, I’d argue that Fry is more interested in raising our awareness about the complexity of these algorithms.
This is a math book at its finest—by which I mean it’s a math book with very few equations in it. Lay people often assume a good mathematics book needs a lot of formulas and numbers, and that’s not true. Math isn’t formulas (that’s engineering—sorry not sorry). Math is about developing a system for solving problems creatively. Fry breaks down what an algorithm is in simple terms, and I loved the chapters on Medicine and Cars, because Fry uses these to explain some great statistical concepts: false positives and negatives, in the former; and Bayesian inference, in the latter. So even though a lot of the anecdotes, specific algorithm examples, etc., were already familiar to me, I still enjoyed how Fry tackles these fundamental but often overlooked mathematical ideas. (As a fan of graph theory and decision math, I also liked the discussion of random forests.)
Fry spends a lot of time discussing how algorithms can get thing wrong. She points out (perhaps obviously) that algorithms will never have “human” judgment—algorithms can’t be empathetic or sympathetic. She illustrates how an algorithm is always going to be biased, so we should be less concerned with chasing after “objective” algorithms but instead focus on building algorithms that are more honest about their biases. The problem with machine learning is two-fold: it’s the data sets we feed in, but it’s also the fact that the decision-making that leads to the output is often opaque.
For all that Fry paints a dire picture, though, she presents a balanced viewpoint that also endorses algorithms as potentially beneficial and necessary. In the Medicine chapter, she points out that algorithmic recognition of diseases like breast cancer is going to make the healthcare system more efficient—as long as these tools are used in conjunction with human judgment, not as a replacement for it. These sentiments are echoed in every chapter, from her exploration of the justice system to her explication of driverless cars, repeated once more at the end of the book where she mentions Kasparov’s centaur chess. If Fry is correct, then perhaps our optimal future is a cyborg future: one in which algorithms enhance our decision-making and help defuse the fallibility of our human judgment, but where humans remain in control of the ultimate decision process and can audit the algorithm.
Hello World is a clear, easy to follow discussion of an extremely relevant topic in today’s society. If, like me, you’re well-read on this subject already, there isn’t a lot of new stuff in here—but I suspect you’ll probably find something. Even so, you’ll hopefully appreciate Fry’s talent for writing and explaining these ideas. As for anyone who has only recently become interested in this subject, you’ll not find many books that explain these ideas so well. Like I said above, this pairs nicely with Weapons of Math Destruction—read both!