How are computers and similar machines so fucking ‘inhumanly’ accurate?

203 views

How are computers and similar machines so fucking ‘inhumanly’ accurate?

In: Technology

The computers were built by thousands of humans. Without pride the computers accepted all input that was given to them. Computers listened to their elders and weren’t stubborn, now they’re the superior beings.

A brain is an analog system designed to deal with the analog world, it’s imprecise, but it can do a lot of things very fast with a “good-enough” degree of accuracy. This system is great for when you need to make very rapid movements in a life and death fight with a bear. It’s a lot less useful when you’re trying to do mathematical equations where there is one exact right answer and you need to be 100% correct 100% of the time.

Computers are the opposite. They’re very rigid, only able to do certain tasks, but they do them with pinpoint accuracy every single time. The reason they can do that is because at the core of a computer, everything is hard-wired in a certain logical way. So if you think of a human brain, you might think of electrical signals flying every which way, causing all kinds of spontaneous connections of different strengths. In a computer, an electrical input of a certain kind will ALWAYS and ONLY produce a certain output.

It’s a logic thing, A goes to B goes to C. B can never be skipped, A will ALWAYS go to B and B will ALWAYS go to C, because it’s a chain reaction built in wire and silicon. So as long as there are no manufacturing faults, and as long as the software is providing the correct inputs, the outputs cannot be wrong.

In terms of why it’s inhuman: because humans are horribly inaccurate. We, as brains, are neural networks that learn through experience. All the math that we know and do is something we had to figure out over the course of our lifetimes. Computers, on the other hand, have the advantage of being “born” knowing – the math that humans took centuries figuring out can be pre-programmed into them. They have no uncertainty (technically not true but for all intents and purposes that’s what we’ll go with) and no doubts about the calculations they perform. The way they work is just different on the ground level, since they aren’t (with the exception of AI) actually learning anything. All they’re doing is taking inputs and following orders.

Humans are solving harder problems than computers and they’re doing so with less energy consumption.

To accomplish this, humans don’t use *exact* logic, instead relying on a series of fuzzy and imprecise heuristics and rules designed to *approximate* the correct solution in a scenario of limited information, limited processing time, limited hardware size, limited energy availability, etc.

If you make computers solve similarly hard problems, they necessarily get similarly imprecise. The only reason computers can solve *some* problems more accurately than humans can is because they’re either easier problems to solve or because the computers are using billions of times as much energy in doing so.