The automation paradox and demand

The Automation Paradox or the Automation Myth asks why, despite many claims to the contrary, has automation yet to create widespread unemployment? Two things caused me to return to this question in recent months. First, the public beta of the ChatGPT, the text synthesis app from OpenAI. Second, finishing James Bessen’s fantastic book, The New Goliaths.

The launch of ChatGPT followed many predictable claims that whole industries would soon be devoid of human labour. Ethan Mollick, was then quick to remind people that only one job from the 1950 US census has been eradicated by automation and that was the elevator operator. In reality, novel inventions such text synthesis augment human labour and make those individuals, and the companies they work for, more productive.

So why no new found leisure time or mass unemployment? If companies can maintain output at reduced costs, that would seem like a pretty straight line to worker layoffs. Well here James Bessen’s timely new book provides great insight. As Bessen explains, the piece that is so often missing is an appreciation of what reduced costs do to demand. If the addressable market expands when price drops, the new productive capacity doesn’t result in layoffs, it results in new customers. In economic terms, greater automation expands markets that are price elastic.

Consider a SaaS startup who are able to integrate ChatGPT or some equivalent tool into their customer support function. The tool makes each worker more productive which enables the startup to lower their prices. The lowered price results in the startup gaining more customers and the additional capacity each customer support agent has gained through the use of text synthesis is then utilised by those new customers. In this scenario, layoffs are irrational, as you need the same number of customer service agents to support your increased customer base.

This is by no means a conclusive answer to the automation myth. Many industries are price inelastic (changes in price don’t alter demand) and so by definition they will exhibit different behaviour. And as Matthew Yglesias pointed out in his long piece for Vox, part of the problem developed nations face is that robots are not taking / augmenting our work fast enough. Meaning, the very notion that the new invention we see can result in layoffs is false.

That said, understanding how automation and demand can work in concert was revelatory for my understanding of the topic, and is just one of many great reasons to read Bessen’s new book.

NB: If you’re interested you can read my quick review of The New Goliaths over on Literal.