What is an Algorithm?
Description:
You have probably heard of an Algorithm and wondered what it is and how it relates to artificial intelligence. This module provides insight into what Algorithms are, how they work, and how algorithms impact everyday life.
What is an algorithm? Video Transcript
Have you ever wondered how your favorite online shopping sites suggest the perfect items you didn't even know you needed or wanted? Or what about your GPS system? How did it know you wanted to go home or even how to get there? Sometimes it can feel like the internet is cussing information based on your own needs.
How does that happen? How does the internet make things so easy for you to find things specifically for you? The answer can be summed up in one word, algorithm. Today we're going to talk about the top three things you should know about an algorithm.
Number one, what is an algorithm? There's a good chance you've heard the word before. Maybe you've heard that computers use algorithms. Or maybe you've heard someone say, the algorithm on that site really knows what I like. Okay, let's start with artificial intelligence, or AI, and machine learning. Artificial intelligence, AI, is the simulation of human intelligence by computers. Basically, AI mimics human thinking at warp speed with massive amounts of available data on the internet.
AI does what the human brain cannot do. AI instructs the algorithm to use really large quantities of data very quickly and to learn from it. This is called machine learning. Okay, imagine walking into a bakery. You walk in, you see all your favorite pastries, so many varieties, so many options. It took recipes to make all of your favorites.
When you are browsing the internet, you see the result of a lot of work that took place behind the scenes. Algorithms are the recipes at work behind the internet. Everything on the internet and many things around us work because of algorithms. Basically, the computer is learning from our clicks and behaviors.
Algorithms apply what it learns from your clicks to get better at what it's intended to do. For example, a shopping site would like you to buy more products, so the algorithm instructs the computer to offer you more products based on what you like. It's pretty simple. An AI algorithm is a set of instructions, a step by step guide or recipe written specifically for a computer. The computer uses this data to learn and improve its performance over time. Well, an algorithm looks something like this. Pretty complicated.
So how does it work? An algorithm instructs the website how to respond to our clicks or how to interpret information or how to guide us to another source of information. Let's go deeper. An algorithm is written by a human being for a computer to follow. You encounter algorithms every day, online or offline.
An example. Algorithms are used to program traffic lights while you're out driving. How? a computer determines how long a light should be, red or green, by using real time traffic data. Similarly, your car's GPS uses algorithms to find the best route for you. When you do things online, algorithms and models are learning about you to improve your experience online.
Algorithms can be complex, such as when your doctor searches for the best approach to help with cardiovascular disease. Doctors are relying on complex algorithms in the background to guide them to the best solution. For example, they can access massive amounts of information about other people who had cardiovascular disease.
In the past, your doctors could only rely on the notes in their files. Not so much anymore. Pretty neat, right? It sure is. Algorithms are necessary part of our experiences online and offline everyday life.
So what can go wrong with algorithms? Unfortunately, algorithms often do as much harm as they do good. Let me explain. When a person writes an algorithm, they sometimes inject their own unconscious bias into this recipe, which is known as code. In other words, a programmer, the person writing the code, might accidentally transfer some of his or her own prejudices into the code itself. For example, when a black woman attempted to use computer vision software called facial recognition that was supposed to track her face, she found that it didn't work.
The software would not recognize her face. She decided to wear a white mask to see if the AI would work, and it did. The algorithm instructions recognized a face when it was white, but would not recognize a black face. That reflects unconscious bias. The person who wrote the code for this facial recognition AI tool did not include instructions to recognize different color faces.
Why is that a problem? Well, think about it. Have you been seeing more and more things happening with facial recognition? What if the tool didn't recognize your face? What if you were trying to access healthcare or something that could save a life? You might be asking, how is this even possible? Aren't instructions for computers plain and simple?
Well, here's how it happens. Unfortunately today, many of these programmers, or the people writing the recipes, are not diverse. As a result, the algorithms may only reflect one perspective and may discriminate against others like women, people of color, disabled, and others. Sometimes the algorithm did not require the computer to consider all types of people in all types of circumstances.
This can lead to a lot of unintended consequences impacting marginalized communities. It is imperative that we address the issue of unconscious bias in algorithms. This will happen when we have more perspectives from different people writing these algorithms. The good news is that more and more technology companies are working on creating more inclusive AI to avoid unconscious bias.