Updated: May 19, 2020
A neural network is machine learning.
What is machine learning?
Math that can learn.
In a technically simple sense, its math in which every output is affected by the previous inputs.
In a general sense, it is “wouldn’t it be cool if there was a computer that could answer any question I asked it without me having to do anything, I just ask and it answers?”
In a broader sense, it allows for self-driving cars, automated restaurants, and robots.
On a national security level they allow data analytics like never before, we can finally ask a computer to search through video for objects, people, and events.
So what is a neural network?
A change in life as we know it.
While it is highly important to know about deep data, statistical mechanics, or any of the technical stuff, all of that can be googled.
But those are how those are why; what is something completely different.
Let’s examine the combustible engine, we can get into entropy, valving, and other mechanics however none of those reveal that the combustible engine can get you from point A to point B.
The math or the mechanics of something never tell you what an object is, just how it works, and this is an exploration of what a neural network is not how one works.
Continuing with the idea of the combustible engine, we find it in the form of a car, and in that form, it transformed the whole of society.
Not only did it shorten the amount of time it took to travel and it transformed cities, but it also caused a sexual revolution, in a way the car is a mobile bed.
And in a way the car is the single greatest tool ever created to terraform a planet, it was hard to see the impact it would have on the global environment.
And like the car, neural networks have implications far beyond just being computational devices.
We’ve already seen what computational devices have done to our lives, and like their underlying mechanics, nothing about ones and zeros can explain how their utilization has changed the world.
It’s hard to predict exactly the endgame effects of neural networks, however, in the short term, the changes we will see will be fundamental.
Firstly you no longer will be able to hold viewpoints that are not rooted or grounded in scientific, historical, and academic data, the answers the ability to understand why those answers are relevant will be available for the current level of laziness people employ in their “opinions.”
It is not a stupider world but a smarter world neural networks usher in.
This is because the type of inputs used for neural networks are more natural to how humans think and ask questions.
Currently typing into a search engine “multiply the length of the Nile river by Herschel Walker’s senior year rushing stats divide by George Bush’s height add the number of words in this sentence” can not be answered, however, in the neural network universe, the answer is easy.
While the previous question may sound strange and not a good example, what it points out is the level of abstraction that can be handled by neural networks.
This means asking a computer to “Make reservations at the restaurant Jill talked about yesterday while we are on the phone for 6 pm,” is computationally possible.
Neural networks need data; the more data the better, which means in the best neural networks the ones that can track what you have said to your friends and be able to apply that data to future inquiries means your life is always being monitored.
To ask a machine to, “Go to the store talk to Jim, Jim is going to give you a pizza, you give him this 20 and say thank you” means that you, Jim, and the robot are all data points in a neural network being continually computed, constantly updated.
Now while it is the first initial reaction to think this is bad I want you to think again; is it only because you are afraid of another human using your data that makes you uneasy about being “tracked” or “monitored?”
If that is the case all you need to do is just employ your own neural network that is specifically trained to alert you when your data is being accessed.
Neural networks will allow you to become unhackable, any time your data moves, is accessed, shows up your own personal neural network will be able to find that data and understand who, what and why.
At the end of the day, neural networks are one of many new tools humans have developed that are in use and already changing the world, it is only when we pay attention to these types of tools that we can avoid employing them in a way that makes life worse for ourselves.
So, for now, a neural network is something that needs to be thought about, it’s something that needs to be talked about because once we start there is no going back and power will rest with those with machines who watch and learn, and if we are to have a wonderful future we have to focus on how to build them and distribute them.