Artificial Intelligence (AI) Definition

This post is part of my speak back in 10 June 2020. I will describe the chunk of this session bit by bit in the next few articles. If you want to check the deck directly, you can download it below. This part will focus on the definition of AI itself which somehow getting popular in recent years. Read on for more details.

Let’s start with an agreement here. Machines don’t learn. I repeat one more time. Machines don’t learn. This is fundamental to grasp to avoid ourself become the hype chaser. Because it’s not practical to follow every single hype especially in technology. I’m not saying we don’t have to learn new things, of course we need to know the advancements of many things. What I want to emphasize is don’t make any new hype as silver bullet. It won’t. Period. AI, as many other great human inventions in technology, is a tool that help us make better and faster decision. The decision maker is still us, human. And we are the one who decide if any machine-made decision is good to help advance the business.

Computers (combination of software and hardware) are generally great at doing two things: performing complex calculations and doing repetitive tasks. For example, if I ask you what is 8,765 times 1,234 without using any calculator or Google, normal human will need some time to calculate this. But for computers, they can do it in a split second. Second example, if we were asked to write a sentence for 100 times, at some points we may lost count, our hands start to get tired, and this can lead to error. Computers generally don’t feel tired or lost track if we asked them to do same thing over and over again.

These two powers of computers are exploited much further in a way that we start to market the terms that they have some kind of ‘intelligence’. To make it sounds fancy, marketing folks coined the term ‘artificial’ to show that we somehow successfully created this. If we want to take a step back, the way computers ‘learn’ is totally different with us. If we have seen a car in a photo before, we can still recognize it although the photo is tilted a bit by certain degrees. For computers, we need to explicitly ‘train’ them to understand this tilted effect. If not, then computers will fail to recognize the car.

As always, human wants quick gratification so this term gained more popularity. So AI now is associated with computers with these following abilities:

1. Perceiving
2. Reasoning
3. Learning
4. Interact with environment
5. Problem solving
6. Creativity (to some extent)

Now at the very core, computers with their strengths basically can try to find a ‘pattern’ from the data we fed them. So if somehow our data can’t provide it, computers won’t ‘learn’ anything. In contrast, if we can provide a wide range of data for a particular task, then computers will definitely help to extract this for us.

I hope this short post will help us understand the basic definition of AI and can guide us differently if we want to know more about this area. In my opinion, AI is really powerful to help us make decision, especially for companies who never relied on their data to make decision, but it’s not a silver bullet. This means human will still be the ultimate decision maker and not vice versa. See you in my next post!