A.I. Isn’t the All-Knowing Superbrain You Think it Is.
If you were to compare A.I. to either a big computer filled with answers or a newborn baby, the latter would be way more comparable.
You see, AI (Artificial Intelligence) is just as its name makes you suspect….intelligence, but artificial. But instead of “it” already being super intelligent, it’s actually pretty darn stupid on itself. The only thing that makes it smart or intelligent is the fact that, once it’s fed with the right information, there is potentially no stopping it.
You don’t code it, you train it
Artificial Intelligence is not a being in itself. It’s the intelligence we can assign to a machine or other form of technology and make it “smart”. Rather than programming it with code which has a beginning and an end, we train it and feed it with data. By learning patterns, the machine is able to understand the outcome the user is seeking and add predictions to the outcome. The more historical data the artificial brain has, the better it gets at recognizing patterns and predicting the outcome of what it has been analyzing. Just like humans: we are not coded. We learn by seeing/hearing/feeling/experiencing.
The first time you saw a dog, your mum or dad probably said:
“Dog….that’s a dog”
You observed and tried to repeat: “Dog”
The next few times you saw a dog, an elder told you:
“Remember…it’s a dooooog. A Dog”
Until finally you saw a dog and said:
“Daddy….dog”
It’s how your brain works. You discover something new and an outsider with “historical data” tells you what it is. In this case: a dog. Your brain then identifies visual anchors, such as 4 legs, fur, a tail, 2 eyes, makes a sound…must be a dog.
But then you see a cat and you say: “Daddy…dog” (4 legs, fur, 2 eyes, makes a sound). Your dad glances at the 4 legged animal and says: “no, that’s a cat”. Your brain starts doing the same thing, but now with more detail. Your brain now starts to understand the difference between a dog and a cat, because it has historical data. It has confirmed what a dog is and sees that a cat has different features (patterns) and saves it in its memory as “cat / not dog”.
The past decided the future
For AI it’s pretty straight forward. The quality and quantity of historical data determine the future of the outcome. The more (clear) information/data it has, the better it will become at seeing patterns and making predictions for the future.
If you want Artificial Intelligence to compete in a game of UNO, you need to feed it with as many previous games as possible. It will then look at all the moves and outcomes of those moves to search for patterns. Once it has a clear indication of how the game is won, it will be able to compete. Once it wins, it will tell itself “success was achieved” and store that game with all others and pull another analysis. This also works the same as with us humans. The more complex something is, the more repetition we need in order to understand it. The better we understand (again, by repeating) the better we will become at steering the outcome.
If you want AI to identify a shower, you need to feed it with 1000’s of images of bathrooms and tell it what a shower is. It will look at the features of showers. At first, only “clearly portrayed” showers will be marked as such. The more features it can recognize, the more diverse the images can become. First, it will only mark a shower when it’s up close and standalone. The more images it has seen, the more complex the environment may become until the AI can spot a shower in a very crowded image with a shower in the background.
AI is not a superbrain, it’s a newborn baby
At this time and age, Artificial Intelligence is not very advanced, in the sense of comparison to where it could eventually go. There is no central computer running an Artificial Intelligence program that’s capable of replacing humans in the blink of an eye. Much rather, we are starting to understand how AI can enhance our lives and help us become more human and more capable. People are working on AI in different kinds of fields. Some are building AI-powered tools to spot medical conditions sooner, others are working to improve customer experience or speed up commercial processes. Like I said, Artificial Intelligence is still in its infancy, but its potential is huge.
