If only it was that simple.

AI will learn faster than humans retrain — total job elimination is inevitable


Unlike the industrial revolution — that led only to a shift in job type — we have reached a tipping point, now, where technology will not only eliminate jobs, but will learn to do the new jobs we are creating, faster than we can retrain humans.

As there will be nothing a human can do better than technology, there will be no human jobs.

Technology is eliminating human jobs

Truck drivers will be the first to go — there are 3.5 million truck drivers in the US alone. And there are already fully autonomous trucks. On some measures, this is the most common job in a majority of US states.

Other drivers will be next, there are autonomous cars are already serving real passengers.

Journalism, warehouse work and customer service are all being eliminated… in fact, a recent report by McKinsey suggests that using existing technology, 45% of all job activities currently carried out globally, could be automated. With existing technology. 45% of all work. Can be. Eliminated.

The pace of elimination will skyrocket

The reality is, only a tiny number of the jobs mentioned above have been eliminated. But the rate of elimination will accelerate exponentially — practically all technologies are rolled out slowly at first, before becoming mass-market. This process is also accelerating.

Furthermore, as generalised abilities of AI increase, they will be able to adapt. To be clear: the solutions we create to problems we are solving now, will be transferable. For a simple example, look at DeepMind’s AI that is capable of learning to overcome a diverse range of challenging tasks: given just the score and pixels, it reaches human-level performance in 49 different Atari games.

And this all matters, because that will increase the rate of elimination — which is precisely the issue.

There is a skills gap

The standard answer to the claim technology is destroying jobs, is that — no — it is simultaneously creating new ones. Or the labour force shifts into jobs that were unfilled.

The issue is that the new jobs require different skills to the old: truck drivers need to become software engineers. But there is already a shortage of talent available to fill skilled job openings. This is a problem we are already not good at solving, and it’s going to get worse.

It means as we eliminate the jobs of humans, they will need to undertake extensive retraining to perform the tasks of the new jobs created.

Training humans takes time & money (we don’t have)

But the systems are not in place to begin training these millions of people. Nowhere in the world. What we do have are monolithic education systems that turn children into degree-waving, educated adults — in 25 years.

In contrast, technology can learn unbelievably fast — and it will get faster. Technology will be able to teach itself. But even if you assume that it won’t be able to, humans will find it easier to teach technology to perform a function, than to retrain humans to do it. Why train 1 million humans when you can write one algorithm?

It will take a long time to retrain a truck driver (which requires no formal education as a prerequisite), to become a software engineer. And though not a truck driver, I’ve first hand experience of learning to code from scratch. I had huge advantages to pick it up — a scientific background with a large amount of logic at degree level.

It is still bloody hard. And most will not have the advantages I had.

We have failed to retrain in the past

Recent history also reiterates the point, demonstrating how we have not been able to build systems capable of mass retraining. UK mining communities are still reeling — 30 years on — from the job losses following the mining strikes. A report from the Centre for Regional Economic and Social Research puts it well:

“the miners’ strike of 1984/5 may now be receding into history but the job losses that followed in its wake are still part of the everyday economic reality of most mining communities. The consequences are still all too visible in statistics on jobs, unemployment, benefits and health.”

And this is 30 years on. In 30 years from now, almost every job activity currently on the planet will be performed by technology.

Technology will replace jobs, and humans will not be able to retrain fast enough.

Anything we can do, technology can do better

Whether it’s our previous examples, winning Jeopardy!, Go (in an entirely new — one might say creative — fashion), being better at diagnosis than doctors, understanding the meaning of data to build models, replacing chefs, reading emotions or detecting lying better than us — you name it, technology will be able to do it.

Every job is vulnerable. And any new jobs created on top of this replacement will be taken by technology too.

The end of the world as we know it

Highly adaptive technology alongside sluggish humans will lead to total human job elimination.

But even before we get there, as unemployment rates in major economies like the US — whose unemployment rate has not been above 11% in 50 years — rise, there will be radical changes that we currently have no functioning solution to.

And when technology does eliminate all jobs, humanity will have to rethink the purpose of it’s existence, because for the majority right now, through will or circumstance, our jobs define us.

Thanks for reading. Recommend. Follow. Feedback. There is more to come, and you’ll make it better 👊