One of the most difficult things about being from the future* is that most people think I’m crazy until what I predict actually happens, at which point I look like a visionary in hindsight, which is not a great business model. (“From the future” is my shorthand for “I am working in tech startups”.)
One of my predictions drawing concerned glances from passersby as I walk the streets with my sandwich board is that AI and machine learning automation is going to wipe out a huge number of jobs in the Australian economy, because so many Australian jobs are in rules-based knowledge processing, and AI and ML just do rules-based knowledge processing better, quicker and cheaper than we humans.
Walk around the floors of any office tower in any Australian city looking for clusters of more than 8–10 people with the same job title in the same department. Why? Most of the office towers in Australia are occupied by service firm of some kind and in these companies, any group of people larger than an executive team (hence 8–10) and doing the same thing (hence same job title and department) are engaged in rules-based information processing.
Approving mortgage applications, checking insurance forms, evaluating tenders, setting up new mobile phone accounts, booking flights and hotel rooms, qualifying sales prospects, or looking for greater efficiency or checking the legal compliance or collecting the monthly reports from those same processes…
All of those processes can be done faster, better and cheaper by software automation. Are being done already but “the future is already here — it’s just not very evenly distributed”. Yet.
Every month another business news story increases an analyst’s estimate of how many jobs in law, accounting, banking, insurance, telecommunications, travel, retail, marketing, and sales will be replaced by software automation. Today, Macquarie Private wealth says it could replace one in four jobs in Australian banking.
So maybe, like me, you work in tech startups and not for an Australian bank, and you don’t need to worry (yet) about software automation replacing you.
But maybe, unlike me, you do work in a rules-based information processing role. You’re not alone.
Get a map of Sydney, Melbourne, Brisbane or Perth. Draw a circle around a 30km radius of the city centre. Colour in (if you’ve got the time) the proportion of the residential housing within that radius which was bought with a mortgage funded by two people, both employed in a rules-based information processing role for a bank. Now colour in those working for a telecoms company. Now add the insurers. Now the airlines. Now the mining companies. Now the professional services firms, such as accounting, legal and consulting.
Have you coloured in nearly all of your circle yet? I have.
Now, when both those people have a one in four chance of losing their job and aren’t able to find the same kind of work elsewhere due to automation, how many have to default on their mortgage? What percentage of those mortgages can be serviced by only one of those two salaries if the household is fortunate enough to lose only one salary to automation?
What does that do to house and apartment prices? If I were you I would find a way to not be left with a mortgage debt worth more than your property by, say, 2022.
This article was originally published on Medium.