Artificial Intelligence — Explain to me like I am 5

Shaikh Abuzar
4 min readApr 3, 2021

Artificial Intelligence, is a term that we have heard a lot, so much that nowadays a tech gadget without it feels like an ancient relic. But have you ever wonder though a lot of companies claim their device has this AI feature, or that what exactly does AI means. If you have and found the answer that’s amazing but if you have not let’s understand it like a 5-year-old.

Before understanding Artificial Intelligence as a whole let us understand what this individual term Artificial and Intelligence the individual term means.

Artificial — It is something that is produced or made by humans which does not occur naturally. Basically, something you create that is a replication and /or enhanced version of a naturally occurring object, substance, etc. A good example would be Stainless Steel, which is an enhanced version of steel.

Intelligence — Intelligence means the ability to acquire knowledge and skills, and apply them. A very simple example is the first time a baby learns to walk, it basically acquires the knowledge of balancing and motion and then uses them to walk (the baby does not know it). Or when you first learned to draw a circle with crayon, you then draw various fruits that resemble the shape of a circle.

Now, what does Artificial Intelligence together means, you would ask. To put it in simple terms it means “Something (typically a machine, or a software) which is created by humans which possesses the ability to acquire knowledge and apply it”. Now AI is a broad category in itself but it can be understood easily if we can break it into subcategories, the following image explains it pretty well.

Image Credits: https://www.aiiottalk.com/types-of-artificial-intelligence-details-that-everyone-should-know/
Image Credits: https://www.aiiottalk.com/types-of-artificial-intelligence-details-that-everyone-should-know/

So now we can say that AI is a human brain that is created by humans to surpass human limitations. But now the question arises that will AI take over the world as they show in Sci-fi movies. Honestly, it entirely depends upon the perception of how you treat AI agents (here agents are referred to as individual AI being), you can see them as a source to cure diseases that are incurable right now, but at the same time you can see them as the source to create a world where humans will be enslaved. Everything about AI is how you perceive it. It depends on how you train it, similar to how you raise a kid.

Now to address one last question has humans today created true AI, which can think and enact like humans, to a certain extent Yes, but there is still a lack of computational efficiency, and replicating the exact human behavior has not yet been accomplished, to accomplish it the AI agent should be raised like a human child, which is as of right now a bottleneck because of the limitation of technology of the present.

Building an AI is like putting the pieces of a puzzle (different types of technologies) together, then raising the AI (training it), and finally letting it explore the world (Testing it on the unseen dataset).

This might be a little out of context but there are two major approaches in developing an AI.
1. Top-Down Approach: In this approach, the AI is given some dataset, then trained on it, and after that testing is done on it. It is similar to how you show a kid a circle looks like (dataset), then help him draw a rough circle by letting him trace on an image of a circle (training), and then ask him to draw the circle on a blank piece of paper (testing).
2. Bottom-Up Approach: You let an AI agent experience things on its own without providing a little to no dataset and allowing it to apply the learned knowledge to try on new things. A good example of this would be how you teach a kid alphabets and then ask them to write new words on their own by dictating the word to them.

Hope you got the gist of AI on a higher level, to learn more you can search the topics individually and study them, but AI as a whole is pretty hard to understand all at once, my suggestion would be to take a topic and understand it, and then proceed to the next topic.

Feel free to provide any feedback and let me know if I am wrong somewhere.
Connect with me on Linkedin.

--

--

Shaikh Abuzar

Just a ML Enthusiast who wishes to share his knowledge and gain knowledge from others.