🤖 What is Artificial Intelligence?

Issue 15 - Technology

Artificial Intelligence (AI) is a term that’s been heard a lot lately. It’s usually defined as the next-generation technology that thinks and acts like humans. However, that’s just on the surface. There’s much more to know about AI, which we’ll be exploring in today’s article!

What is Artificial Intelligence?

Before we get into the meat, we need to take a look at how AI came to be a thing. In 1955, a Computer Scientist named John McCarthy at MIT (Massachusetts Institute of Technology) invented the concept of “Artificial Intelligence”. He also created the first AI workshop at MIT but later transferred to Stanford along with John. He is considered “the father of AI” after presenting the concept at a conference

AI can be generally defined as a machine that is programmed to simulate the behavior and thought process of a human. There are 2 types of AI: Strong AI and Weak AI. Strong AI is when you have a machine that shows behavior that you’d expect from a normal human. Weak AI is a limited and programmed system to do narrow tasks. It’s trying to assist you, but not more than that.

Here’s an example: When you ask Google or Alexa - “How are you doing”, they will respond, “I’m fine”. This is Weak Artificial Intelligence because they are programmed to do this. They may have different responses like “I’m pretty good”, but here the system is choosing from a list of options in the database. So, the machine is just matching the input (“How are you doing”) to an output (“I’m fine”).

It has to be noted that, most of the AI we experience today is Weak AI. Strong AI is still more of a Science Fiction at this point. Therefore it’s believed that the AI we are presented with is just the tip of the iceberg, and there is still a lot more to come to advance the technology.

Machine and Deep Learning

Machine Learning (ML) and Deep Learning (DL) are a pair of terms that appear to be the same, but these are 2 sub-categories under Artificial Intelligence. Both Machine Learning’s and Deep Learning’s end goals are the same, which is to build an artificially intelligent system.

Both of these concepts came from a similar idea: Instead of experts teaching machines to do things, what if machines could learn things by themselves? This is how Machine Learning and Deep Learning came into existence. The experts create symbolic patterns and have the computer learn from absorbing them.

With Machine Learning, you need to give a huge sample size to the system for it to observe, analyze, and come up with a good picture of what it is. Deep Learning is the exact opposite approach. You feed the machine a small sample size, and the machine will analyze the samples “deeply” to get the maximum amount of data, and produce results.

Here are a few examples. The first one is about Machine Learning, and the second is about comparing the two learning approaches.

(1) ML - Siri may track your location and time throughout the day, so that when you get in the car in the morning, Siri may say something like “Do you want to navigate to work” or “It will take 15 minutes to get to work”. Siri is learning your location and time and therefore coming up with these suggestions based on that, therefore this is ML.

(2) Comparison between ML and DL - Let’s say there’s a program that can tell the difference between dogs and cats. With an ML-based system, you would need the system to analyze things like the ears, nose, overall shape, etc. It would have to analyze a lot of photos to tell if the animal is a dog or a cat. With a Deep Learning system, you would feed the machine one picture. It would absorb everything that it could from that picture and then be able to tell the difference between the 2 animals.

Unfortunately, systems that use Deep Learning aren’t available to the general public yet, because it’s very sophisticated and the computer would have to squeeze every possible detail out of something. On the other hand, ML is so much easier to make because the computer just has to analyze lots of data to come up with something.

---

By the way, if you are enjoying this article, please be sure to share and subscribe. I have a promotion going on which allows new and existing subscribers to get a 1-month free trial to Skillshare (the best place to learn any new skill)

---

Neural Network

Neural Networks are another subset of AI. Neural Networks are based on the structure of our brain. A biological brain has billions of neurons in it. They connect and send signals as a way to learn and react to the world around us, and they communicate using synapses. Synapses allow neurons to connect and communicate with other neurons in the network. The neurons will increase the strength of their connections based on your experiences.

In a neural network, the neurons are organized into layers. There’s the input layer, the hidden layers, and the output layer. A good comparison would be a marching band. For example, the band is given a note, and they are told to recognize it. This would start with the input layer, which is the band leaders. The Band Leaders, take note and make the people in the hidden layers (the players) recreate the note. They would tune their instruments, by trial and error, until they get the desired outcome.

The idea behind the Neural network is that if you passed the input through the hidden layers enough times, you’d end up with a complete picture, in this case, the sound (narrowing down possible options). The only challenge with this approach is that it can be time-consuming. This is because this is just trial and error but on a very large scale. As a result of this, experts will try to tweak the network to make it a little more efficient.

Supervised and Unsupervised Learning

So after building an Artificially Intelligent System, you need some way to teach it to recognize patterns. The machine doesn’t just match up x to y. Instead, it recognizes the patterns going on in the data and learns them. However, you need some way of getting this data to the system. One option is Supervised Learning. This is where you give the machine a small set of data (training set) and the machine will process that data, along with the help of a data scientist. Another way to learn is Unsupervised Learning, which is the opposite of Supervised Learning. In this type, the system is given a data set to analyze without any help.

Thanks for reading this week’s article and I hope you learned something new about Artificial Intelligence! If you have any questions or comments, feel free to send them down below. See you next week!

-Luke Rapaka