Work is What You Make of AI

Last month I got all on a kick about how automation is going to take our jobs and that that is a good thing.  There has been some more press on this topic so I want to continue kicking it around.

The first article for my shouting-into-an-empty-house discussion is from the BBC: How long will it take for your job to be automated.  This is about an Oxford study that asked 352 scientists how long various jobs would take to be automated.  This is a better information than the one that the Will Robots Take My Job site used.  And it offers up a time frame that is more heartening for those that are hip deep in the status quo: 120 year before 50% of all jobs are automated.  Hold on to that thought.

The second article is about Eric Schmidt, Google Founder and Techno Spouter (of course I'm jealous, people ask his opinions instead of me trying to force them down any ear that will listen).  In it, Mr. Schmidt puts forth the idea that A.I. will create more jobs that can't be filled instead of destroying them.  His reasoning is that automation will make workers more efficient, more productive, but that humans will still need to be part of the process.  He cites a McKinsey study that says that 5% of current jobs can be automated with today's technology for his reasoning.  It's a very together-we-are-stronger statement, but it has a few holes.

What It All Means

On the surface, these two articles seem to be saying that jobs are safe.  That work is safe.  That our current economy and the way we view our self worth are all safe.  I'm going to give it all a big MAYBE.

First, lets have a bit of a think on those scientists giving that 120 year time frame to half of all jobs being automated.  At first glance, that time frame seems too long.  The pace of technology being more exponential than linear, my gut says that this number is maybe twice as high as it should be.  But then, these are smart people that got surveyed.  So, I had to ask myself, "Self, why are they claiming such a long time?"

My answers came on two fronts.  The first is that is all jobs, not your job (or mine).  In the article, there is a spread with many employment sectors getting automated much, much sooner.  Laundry folding, truck driving (transportation in general, IMHO), retail sales will all come much earlier, hitting the 50% mark in 10-15 years.  In fact, if we were betting, there is a fifty percent chance that your job will be automated before 120 years.  That begins to feel closer to right.

But there is another side to all of that.  'Jobs' are created, lost or automated in a that perfect economic world that my college professors all had us imagine with their supply and demand curves.  Instead, there are other forces at play.  One of the biggest is politics.  People get elected by the number and quality of jobs that they bring to their constituents.  Or claim to bring.  Loosing jobs to automation is a big issue in the halls of power and politicians are using it to power fear among the electorate.  This will artificially slow down the shift from manual to automatic in the gear box of labor.

Now, let's add in Mr. Schmidt's comments: AI will create more jobs than it consumes.  What is not said are some of the caveats from the study that he is citing: 5% of current jobs with today's technology.  Neither of these assumptions are static.  What labor is will change and certainly technology will advance, through the application of that same labor.  On a long enough time scale, technology will consume paid labor.

Enough Jobs to Go Around?

I absolutely believe that combining human labor with automation will result in the best possible output.  Where I get concerned is with the current economic assumption that we all must work.  Even, will we all be able to work?  Will there be enough of these combined human/AI jobs to give the current 7.5 billion people the economic power they need to house, clothe and feed themselves?  And by the time we reach that 120 year, 50% cut off that number will not be 7.5 billion but something far north of 10 billion.  I think not.

I'm going to leave the answer to this all up in the air.  Does it mean something like Universal Basic Income?  I think that is part of the equation, but certainly not the whole answer.  Is it human labor hiring quotas?  I hope not as I suspect that will decrease efficiency at a time when we will not be able to afford it.  What it is is way too complex for one lone blogger to figure out on his keyboard.

I do know that now is the time to start acting.  We cannot wait 120 years for 50% unemployment.  Whatever answers we collectively enact now will not be perfect.  They will have unintended consequences.  They will need to be repealed and re-enacted over and over again.  But all of that will need to happen no matter when we start, and it is easier to do those things when there aren't 3.25 Billion people screaming about the food they aren't eating.