“Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoatnt tihng is taht the frist and lsat ltteers be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.” This phrase was developed by researchers at Cambridge University to demonstrate the amazing power that the human brain has. To solve this puzzle, humans have to not only use their skills of recognition, but we also have to have a deep level of intuition based on years of reading and understanding context. The human mind is deeply and incredibly powerful; capable of solving a puzzle like this without second thought. Our brains have a series of neurons that send information to each other that allow us to form thought, consciousness, and learn. Neural networks are a composed of a series of mathematical functions, arranged in such a way that it mimics the human mind. Most computer functions are done by telling the computer to do a specific task; the computer can only perform that task, and it will never learn to do otherwise. However, with neural networks, we teach the computer how to learn to do that task. Today we’re going to dive into the world of neural networks, and learn how they are currently changing our society and drawing us into the Singularity.
We actually interact with neural networks every single day on a small scale. Have you ever found yourself getting frustrated with the autocorrect function on your phone? Maybe you spelled a word wrong once and now your phone constantly suggests that misspelled word? The autocorrect on many phones now are built using neural networks. This means that every word you type and the order in which you type them are constantly being fed into a stream of tiny neurons that record everything. They begin to detect your patterns of speech, the slang you use, and how often you use certain words; to be able to try and predict what you will say next. Some phone keyboards such as SwiftKey are so good at predicting what you will say next; that it can often predict whole sentences that you are likely to say. Another common example of a neural network is in the google photos app, which can categorize photos based on the content within the photo. Say for example you like hiking– the app can literally recognize all of the photos of you hiking and organize them as such. Ever been on Facebook and had the app recognize one of your friends in the photo and recommend you to tag them? This is just another example of a neural network at work.
So what’s the big deal about neural networks? Yes, they can decipher what’s in an image or what you want to say next in a text, but why are they important? Neural Networks are important because of the way that they gain their intelligence. They learn. They are modeled to learn like the human brain learns. As we are seeing neural networks be applied to our world in interesting and unique ways every day, the Singularity is drawing nearer.
If you have ever spent any time around children, you will know that they ask a lot of questions. What’s that? What’s this? This is because they are soaking up as much possible information that they can about the world we live in. When you tell them that a ladder is called “ladder”, or the house is called “house”, it reinforces the connections in their minds, so that the next time they see a ladder or a house, they will know exactly what it is without having to think about it. Humans learn remarkably fast. However, with neural networks it isn’t that easy. We feed neural nets millions of examples of the thing that we are trying to teach it, as human intuition is something that it doesn’t have yet. However, what we have in depth neural networks make up for in breadth. While we can see one picture of a chair and deduce that a La-Z-Boy is in the same general category; a neural net cannot. For this reason, neural nets must be fed millions upon millions of images of a chair in order for it to mimic the human like intuition to know that a La-Z-Boy armrest fits roughly into the same category as a chair.
The process by which neural networks learn is called Deep Learning. When these neural nets are able to be applied to more things in our society; the Singularity will become incredibly close. We will have a vast number of Artificial Intelligences that display human-like intuition for the things in our world which will allow us to understand our world on a deeper level.