Note: this post is part of a series about the foundations for Code Inspector, you can find the previous article here.

Since the industrial revolution (which started about 200 years ago), we are delegating more complex tasks to machines. At first, we delegated simple tasks on an assembly line. These machines were better than humans in all aspects: they do not go on strike, they do not stop to go to the bathrooms but most importantly: their results and performance are consistent and predictable.

Les Temps Modernes

The result of the machine does not depend on the discussion it had with its partner the night before nor how much wine it had at the bar while chatting with its friends. The machine delivers results according to objective, non-ambiguous specifications.

Until recently, machines were dedicated to automate (mostly) blue collar jobs (factory, farming, etc). But in the last twenty/thirty years, machines took white-collar jobs too (in finance, security, sales, marketing). Today, the computer fixes your grammar mistakes and optimizes your finance better than many humans would do. They do it cheaper, faster, accelerating innovation across the world.

Machines start to take over very complex tasks. And they are better than us at it. The best example of how machines improved over the last ten years is the autopilot technology from Tesla. On highway, a Tesla drives better than you would drive. The car stays perfectly within its lanes, better than you would do. It still fails from time to time (especially in construction zones) and adapting the car to such situations is just a question of time: 90% of the functionality is here and achieving full self-driving is just a question of time.

Tesla Self Driving

The most important aspect is that machines do the tasks they are dedicated to better than humans. They do it consistently, predictably and only fail within their Mean Time Between Failure (MTBF) which is measurable and again, predictable.

Are we doomed to be replaced by machines? That is still not clear. Machines are great at performing a task they are programmed for (such as tighten a screw or drive a car). They will get better every day. And they will learn to complete more complex tasks year after year.

But machines still need to be told what to do. We can program them to do perfectly one single task and they will do it better than us on all dimensions. But they are not the entity where ideas originates at first. That part is still dedicated to humans. For now.