Artificial Intelligence: What is it?

Ronit Taleti
5 min readOct 17, 2019

Judgement Day is upon us… or is it?

Artificial Intelligence is a somewhat growing topic of conversation at the moment. AI is becoming better and better, and while the stuff it can do is cool, people are worried about losing their jobs, or worried about immoral people using AI for crime, or worried about a sentient AI being made.

But to me, I believe AI is a boon, a blessing. AI will change our world forever and we will likely have nothing to worry about.

What is AI?

Most people think of AI as something like HAL 9000, or SkyNet, or many other famous examples of autonomous robots in media. But really, an act as simple as making a search on Google or checking your YouTube feed is you interfacing with AI, or at least the result of an AI’s work.

You see, underneath most of what we do on the internet, there are usually AIs calling the shots, whether we know it or not.

When you search for videos of YouTube, an AI gathers data about your searches and learns your preferences. The next time you open your YouTube, it curates the videos in your feed based on your searches.

These are some big companies that use AI in some way, shape or form.

AIs are very good at their job. Here’s a story. In around 2011, an angry man stormed into a Target retail store. His high school daughter had been receiving coupons in the mail for things like cribs and diapers.

To the dad, he believed Target was encouraging a minor to get pregnant. The store apologized, but a couple of days later, heard back from the father. He said that “there were things happening that he was unaware of. My daughter is due in August”.

Target’s AI systems were able to figure out a woman was pregnant, before her own father. They detected changes in her usual purchases, and the new things she was buying were evidence she could be pregnant.

How does AI work?

AI works by taking data and processing it using intelligent algorithms. An algorithm is sort of like a list of rules that help you solve a task. Smarter more intelligent algorithms can also learn from data to solve a problem.

For example, you could have some data with some labels, let's say the size of a dog and what kind of dog each entry is. Different kinds of dogs have different heights. An AI can learn what height each dog is and if you measure a new dog, the AI can use what it’s learned to predict what kind of dog you measured.

There are three main types of AI algorithms.

1. Supervised Learning

Supervised Learning is when an algorithm learns from labeled data. Think about your teacher teaching you how to do a math problem. Basically, it’s like a supervisor provided material, and the kids learning from it.

In supervised learning, there are generally two types of problems.

  • Regression problems which is predicting a value (e.g price of a house)
  • Classification problems which is sorting data (e.g is this mail spam?)
Classification is sorting the data, and regression is fitting the data.

2. Unsupervised Learning

Unsupervised Learning is when an AI is learning from unlabeled data. Think about giving a human random data and asking to classify it. Soon, they would notice similarities in the data and sort it.

Unsupervised learning is generally for sorting very complex data that a human would have a hard time understanding.

This shows how unsupervised learning can divide data into clusters. This works well when you know the number of categories you wish to have.

3. Reinforcement Learning

Reinforcement Learning is sort of like Unsupervised Learning, since its learning off unlabeled data, however, you can associate a positive or negative response based on the algorithm's actions. In our world, it's like learning by trial and error.

It’s good for problems where you receive feedback from the environment (like getting damaged), and that have a reward (incentive to learn).

When the agent makes an action, it gets feedback based on how right or wrong the action was. It learns from its mistakes, gradually becoming better.

AI, Machine Learning, and Deep Learning

First off, let’s learn about subsets of AI and how they are related to each other.

  • Artificial Intelligence is when a program or machine completes a task based on a set of rules, an algorithm.
  • Machine Learning (ML) is a method of training algorithms on a dataset to make accurate predictions about new data.
  • Deep Learning is having the ability to identify patterns and classify types of information. Deep Learning uses Neural Networks to do this

Deep Learning is a subset of Machine Learning, which is a subset of Artificial Intelligence.

AI could be imagined as programming a series of if statements, while a Machine Learning AI could be imagined as having a dataset and learning from it. A Deep Learning AI could be imagined as discovering data or features, rather than being provided it.

Why should we have AI?

AI helps create increased convenience in our day to day life. We already use it in almost everything. Here are some of them:

  • Search: Without AI, your search results wouldn’t be personalized, which would make it harder to find info for that essay you’ve been working on or the movies/music of your choice.
  • Smart Products: We can run simulations of roads where we can teach AI to drive cars. Now, we have companies like Tesla innovating the car industry. We have smart assistants too like Alexa and Siri which can speak to you in real-time.
  • Healthcare: AI is good for sorting patients, or even predicting diseases up to a year before symptoms show.
  • Fraud: AI is now smart enough to detect fraudulent scams that aim to steal your info.
  • Security: Most iterations of malware change slightly, meaning AI has an easy time detecting new malware files.

Key Takeaways

  • AI is a program or machine completing a task based on a set of rules.
  • AI is used in practically everything.
  • Machine learning is a subset of AI where you make predictions on new data based on previous data.
  • Deep Learning is recognizing data and learning without a dataset
  • There are three main types of algorithms; Supervised, Unsupervised, and Reinforcement

If you enjoyed reading this article or have any suggestions or questions, let me know by clapping or commenting. You can find me on LinkedIn for my latest work and updates.



Ronit Taleti

I’m an avid 17-year old blogger interested in new and emerging technologies like Artificial Intelligence, Blockchain, and Virtual/Augmented Reality.