Hannah Fry Black Swan, 2019, PB, 320pp, £8.99, 978-1784163068
I’ll be the first to admit that seeing a mathematical book flop through the letterbox did not leave me jumping for joy. However, I am now slightly ashamed of my initial disheartened reaction. This book is an accessible page turner on mathematical algorithms — something I never thought I’d say. The book explores how algorithms are inherent within modern life, deeply entwined in our relationship with the digital world. As a naïve GP registrar reading this, I was led to new discoveries concerning how private companies collect and harness the data of individuals.
Fry travels through a variety of hot topics in the digital world, covering everything from Facebook’s political endeavours, AI in health care, to cases of false identity in criminal investigations. In one unnerving example, Fry describes a supermarket’s ability to detect the chance of pregnancy from the items in a person’s shopping basket, then use this data to send pregnancy- and baby-related coupons to the customer’s home. Which in one poignant story led to a teen pregnancy disclosure to an unhappy granddad-to-be.
The chapter on medicine describes a competition between pathologists and machines in diagnosing tumours from pathology slides. Remarkably, the algorithm manages to diagnose 92.4% of cancers correctly but at the same time detects a large number of false positives, leading Fry to suggest that we should not be fearing that our jobs will be taken over by machines, but in fact we will be working in a team, collaborating to make more accurate diagnoses.
Later in this chapter the dilemma of digital healthcare records in the NHS is raised, candidly pointing out the ‘mess’ of NHS data, which is common across many other countries, the US included. Fry points out the difference between the chaotic nature of digital healthcare records compared with the meticulously collected private sector data that brokers sell. This is something we will all recognise at work when trying to find out what happened during a hospital admission or find an elusive X-ray result.
I found my eyes opened to the moral and ethical dilemmas posed by algorithms — should a driverless car save those on board or should it plough into a group of pedestrians on a zebra crossing? And yet I also felt reassured. Algorithms are designed and programmed to be our assistants, aiding humans in making decisions and needing human supervision to work effectively.
My previous concern — that an AI machine will take my place in my GP surgery, happily running a morning clinic — is unfounded, and, from all accounts, a very long way off.
- © British Journal of General Practice 2020
