Artificial Intelligence - It's not as out of reach as most people think

Artificial Intelligence, or AI, is a popular buzzword in tech. There are AI firms all over the world. For most people who aren't in the field of AI, it can seem complicated and mysterious, but in fact it really isn't that hard to understand. While you may think AI is out of reach from anyone without a deep math and computer sciences background, you may be surprised to learn that most applications of AI these days don't have anything to do with complex research. In this post I'll go over what AI is, some examples of what it can do, and how you can get started in the field yourself, should you wish to.

What is AI?

Looking at the definition, AI is intelligence demonstrated by machines, unlike the natural intelligence displayed by humans and animals. This is a very broad definition, and as you might expect it includes a lot of different technologies. There are many ways to make computers smarter, and many applications where this can be useful. From text to voice to pictures and even actual actions in the real world.

One popular method is called Machine Learning, or ML. This uses mathematical algorithms to process large quantities of data, called training data, to find commonalities so that decisions or predictions can be made by the computer, without being directly programmed to do so. For example, an algorithm may be developed to look at photos and distinguish between cats, dogs and other animals. This can be useful for a myriad of reasons, from security to marketing. What ML does, is simply take thousands or millions of photos, go through them, and learn the characteristics of all the different animals. Then, once the software has been trained, a new photo can be fed in and it should be able to figure out what animal is in the photo.

Where can AI be found?

In a word: everywhere. While the thought of AI may bring up fully autonomous robots roaming the streets, the truth is much more mundane. Here are just a few examples of AI in action from daily life:

• When you go to a website and a chat popup appears, with an avatar asking you if you have any question, this is typically powered by AI. There's no one waiting in the background for a visitor to chat with you, instead it's powered by software that can read what you type and interpret your questions, answering them to the best of its abilities.
• When you search for a term on Google, and you not only get a list of websites but actual information shown to you, often times based on your own profile, where the company knows the type of searches you do, and what type of information you're most likely looking for. There's a tremendous amount of AI software running behind the scene.
• When you go to your email account and look at that junk folder, you may notice a lot of undesirable messages there, from the obvious spam to hard to detect scams. Chances are you didn't manually dump all these messages in the trash, instead an AI read the message and took the decision that this was likely junk.
• When you apply for a new auto insurance, an AI looks through all the data the company has about you to see whether you're likely or not to get into a car accident. Your rate will then vary based on that result.
  

AI as a service

To come back to the original premise, while a lot of AI may look complex and almost seem like magic, quite often it's more within reach than you may realize. While large tech companies like Google, Amazon and Microsoft spend billions on AI research and have entire teams of smart people to implement that software, a lot of these same companies also make AI services available to others. For example, AWS offers a long and quickly growing list of services related around AI.

 

Let's say you want to setup a service that would need, for some reason, to distinguish dogs and cats in photos. There's no need to gather thousands of images for training, write your own AI software, and spend all that money on research. Instead, for a very low cost, AWS Rekognition will do the work for you, in a single line of code:

aws rekognition detect-labels --image '{"S3Object:{"Bucket":"mybucket","Name":"myphoto.jpg"}}' 

This will return a handy list of everything the AI is able to detect in your photo, allowing you to use AI technologies without ever spending one dime on research. And this is just one of many examples. Chances are, unless you're on the deep edge of AI, there's a service out there that can provide you the type of information you need, and this is the future of where AI is going. Research will of course continue in large tech firms, but then that will trickle down in the form of web APIs, allowing anyone to use AI for their own personal projects for a fraction of the cost.