The Truth About Artificial Intelligence

…and who’s going to lose their job.

There is a counterintuitive, paradoxical theory of usefulness in Artificial Intelligence:

What’s hard is easy, and what’s easy is hard.

That means if it’s hard for humans, then it’s easy for computers. If it’s easy for humans, then it’s hard for computers.

For example, an image recognition AI can identify cats in photos 80% of the time. Your four year old niece can do that 100% successfully.

Your four year old niece can’t play chess, but your computer can. Chess is hard. Cat photos are easy (for humans).

Most white collar jobs nowadays operate in the realm of bits — cyberspace.

Blue collar jobs are based in the world of atoms — aka real life.

White collar jobs mostly require things that are hard for humans and easy for computers: spreadsheets, data analysis, image editing, etc.

Blue collar jobs create value by using physical objects to modify other physical objects.

Think plumber versus coder.

The problem for white collar jobs is that it is cheaper and easier to create in the world of bits than in the world of atoms.

That’s because, in the world of bits, you’re just moving pixels around on a screen.

In the world of atoms, you have to move metal and stone.

Moving bits is cheaper and easier than moving metal.

That’s why white collar jobs will be the first and most affected by Artificial Intelligence.


The quintessential example of AI is self-driving cars.

Self-driving cars are great, but they still require a human at the wheel. They probably will for a long time to come.

That’s for two reasons.

1.) Technological limitations — the self-driving car A.I.’s can already handle 80–90% of most driving, but identifying and handling the last 5% of driving situations is 100x harder. This is according to Elon Musk himself.

2.) Legality and Liability — blue collar industries are exponentially more regulated than the white collar.

(A skeptic might wonder if that’s why virtually all technological innovation in the past 50 years has occurred in the white collar tech industry.)

That means not only will A.I. have to overcome cost and technological hurdles, but then real-life human lawyers will have to create a legal environment for the A.I.’s to be deployed.

You know, come to think of it, a self-driving lawyer might be handy…