The idea in a nutshell : There are some reports, from reputable sources, which suggest a lot of people are going to be made unemployed by artificial intelligence. It makes sense to have a plan on what to do if a lot of people are put out of work.

Introducing the ‘Usual Suspect’ articles to prove this point

I listened to a fantastic podcast ( sent to me by Nick Cordrey – thanks Nick ) which made me laugh out loud. In his presentation, Edgar Whiteley talked about the ‘usual’ stories in the press which relate to AI. See below.

Below : An amalgam of typical headlines associated with Artificial Intelligence in the press.


I laughed when I heard this because, I must confess, I had read articles exactly like the ones he was talking about an incredible number of times. I had formed the majority of my views on the subject from them. Here are the super scary, often repeated forecasts of unemployment associated with AI.

Scary fact Explanation and source
5 million · The BBC showed 5 million jobs (40% of the Australian workforce) face a high probability of being replaced by computers over the next 10 to 15 years.

• The report says there is a high probability that occupations such as accountants, estate agents and economists will not exist or will be significantly depleted by the 2030s.


Of the occupations displaced

• A 2013 paper by Carl Benedikt Frey and Michael Osborne, of the University of Oxford, argued that jobs are at high risk of being automated in 47% of the occupational categories into which work is customarily sorted. ( Source : The Economist : The Oncoming Wave )

Of US job susceptible to automation

• In 2013, researchers at the University of Oxford’s Martin School conducted a detailed study of over 700 job type and came to the conclusion that nearly 50% of US jobs will ultimately be susceptible to full machine automation.

Of jobs will fall victim to automation in 20 years

• In February 2015, a parliamentary report to the UK’s House Of Lords estimated that 35% of UK jobs will fall victim to automation in the next twenty years.

These are concerning numbers from several reliable sources including the BBC. I was convinced.

The difference between Narrow AI and General AI

The threats here relate to narrow AI. Narrow AI can be ‘trained’ to perform a simple task like recognising cancer cells in a picture which may or may not contain them. As we’ve seen, some narrow AIs have been trained to win at Jeopardy. Narrow AI shows itself more often in ways you don’t realise. The product recommendations you see or the search results you are provided.

General AI is something that would pass the Turing Test. Essentially, a computer with general AI is something with ‘life skills’ so good it could pass a human test. For example, knowing how to reply to someone saying: “The Bible says Jonah hated living in a Whale but my nan has lived in Wales for years and she loves it. Why do you think that is ?”

We are a long way from that although some people are rightly concerned about it. General AI is what Skynet was in the Terminator films and it tends to be what people are most scared of when they hear about Artificial Intelligence. Most well informed observers don’t believe we will see it for some time. Maybe 20 – 100 years.

What would you do if you believed these forecasts

I guess my point of view is that it is usually best to have a realistic view of the world, including the threats you might face. One can then make some rational decisions as to what you want to do to either mitigate, avoid or accept the threats.

  • Researching jobs which are more likely to be ‘safe’ : Even a small amount of research in to AI shows some jobs are more likely to be safe than others. Any role involving people doing something very human is more likely to be safe from Narrow AI than another job which doesn’t have that characteristic. A Business Development saleswoman, a job which requires empathy, likability and an ability to subtly influence people ( like my wife ) is likely to keep her job for longer than, say, an accountant. For my part, I went to get a job at a bank which involves strategy and an intuitive step. Even the most advanced AI systems find intuition hard.
  • Saving as much as we can : We ( my wife and I ) have a savings account which we use for a couple of reasons. First is a buffer for any significant life events like losing a job or needing to travel quickly – for example if there was a family bereavement overseas. Secondly, this money will be held in an offset account when we get a mortgage. I like to save as much as I can to buffer, offset and for a rainy day fund. And then, of course, if the worst happens and we find it hard to get back in to work after being displaced, we have something to fall back on.
  • Putting money in to Super plans : Super is a tax efficient way to save for later in life when AI will be even more advanced and jobs might be more scarce.
  • Investigating and practicing living cheaply : I spent most of 2014 living on my boat. I totalled up the bills I had for food, maintenance and staying in marinas. They averaged around $2k a month. With savings and perhaps Super later in life, we should be able to survive on the boat for some time to come.
  • Alternative income flows : It may also make sense to have a secondary business to work on. Internet businesses and affiliates ( link ‘affiliates’ to ) are cheap, portable, can be profitable and insure against a loss of income from a full time job ( if it is displaced. )
  • Buying shares : The economics behind all this is cost cutting and productivity. By automating tasks with AI solutions, businesses displace staff and lower their costs. As I’ve covered elsewhere, the economic benefits could be substantial and would appear to flow towards companies. Buying shares is probably a good idea at this point.

Summing up :

Artificial Intelligence raises the important question of ‘rent’ in the economic sense. Rent is any income derived from a factor of production. When there is no work, the only kind of income we receive is likely to be from an asset. Having Super, savings, shares and even a house to rent out ( while you stay on a boat, for example ) is all useful insurance against the threat of unemployment. It will be no use saying ‘I had no idea this was on it’s way’ when your job is displaced. There won’t be the money around for the government to help you.

While there are some questions over the models which have the concerning claims about the number of people who will be made unemployed, they can’t be ignored. It does seem prudent to be taking steps which consider these risks, even if the net result might not be as bad as we believe.