47% of today’s jobs could be automated in the next two decades.
It popped up in an Economist article on automation, but I have seen it in a few places before.
When Instagram, a popular photo-sharing site, was sold to Facebook for about $1 billion in 2012, it had 30m customers and employed 13 people. Kodak, which filed for bankruptcy a few months earlier, employed 145,000 people in its heyday.
The analogy sounds like an impressively devastating indictment, but Kodak was imploding before smartphones with cameras were even on the scene; let alone Instagram. And the analogy fails on many other levels.
Today companies can quickly pick up value without profits based on a user base. It's a completely different model than the standard corporation and it's not a point of comparison. Even if Kodak and Instagram had overlapped more directly in any meaningful way, the analogy still wouldn't be meaningful.
Kodak made a physical product. You can compare it to HTC, which employs quite a few people, not to Instagram.
The larger point of the Economist article about the dislocation of workers in the service industry may be important considering how much of the American job market now depends on service jobs, but the article doesn't really make a credible claim for this mass automation expansion.
There are a shortage of actual examples besides Kodak and Instagram, instead we get suggestive passages like these.
Services may be even more vulnerable. Computers can already detect intruders in a closed-circuit camera picture more reliably than a human can. By comparing reams of financial or biometric data, they can often diagnose fraud or illness more accurately than any number of accountants or doctors. One recent study by academics at Oxford University suggests that 47% of today’s jobs could be automated in the next two decades.
But what does that really mean? We're not about to replace doctors with machines and accounting software already exists. All that this means is that doctors and accountants will have better software to work with.
And security guards probably aren't going anywhere. A computer may be able to more reliably detect an intruder, but it's 'dumb' and more vulnerable to being fooled once a weakness is found. And unlike Pete the security guard, a computer can't pick up a flashlight and check out a problem.
The Economist article does miss the obvious. Call center operators may become a thing of the past in less than two decades. Getting a human on the phone will become harder than ever. Emailing customer service will tangle you in Turing Tests.
Physical labor jobs these days are not really worth replacing with machines. But a job that requires you to sit in front of a computer is much more vulnerable to being replaced with software.