Automate This

Steiner, a journalist with an engineering background, wrote about the spread of algorithmic decision-making from finance into other fields, published in 2012 before machine learning became the dominant public frame.

The finance sections trace the history: the first electronic order routing, the quantitative funds, the migration of physicists and mathematicians into trading, the growth of high-frequency firms and the arms race in latency, and the 2010 flash crash. He argues that finance served as the proving ground because the feedback was immediate and the money was available to fund the talent.

The extension chapters cover other fields where algorithms were beginning to replace human judgement: music, where a system analysing hit records claimed to predict commercial success; medicine, where diagnostic and imaging systems were being deployed; law, where document review was being automated; hiring, where personality assessment and voice analysis were being applied to candidate screening; journalism, where automated systems generated sports and financial reports from structured data; and customer service, where call routing systems matched callers to representatives by personality type.

His general argument is that any domain where decisions follow from patterns in data will be automated, that the professions have no special protection, and that the effect will be to concentrate returns among those who build the systems.

The book is journalism and predates the deep learning results that changed the technology entirely. Its predictions about which fields would be affected were broadly correct in direction while the mechanism turned out to be different from the rule-based and statistical systems he describes. The finance history is the most solid part. Readers wanting current analysis of automation and work should look at more recent material.