Machine Learning and Artificial Intelligence: What are their differences?
The thing with new technologies and terminologies is that people tend to sum them up together, often mistaking one for the other. This is true with artificial intelligence and machine learning. There are times when people say artificial intelligence when they mean machine learning, and vice versa.
Don’t worry! If you do not fully understand what each technology entails, then you should know that you are not alone. However, if you are planning to use machine learning, artificial intelligence, or both for your business, then you should at least understand the differences between these two in order to make the right decisions.
Artificial Intelligence vs. Machine Learning
Artificial intelligence enables the system to manifest intelligence as if it were a human being. On the other hand, machine learning is a type of artificial intelligence that makes use of algorithms or mathematical models that pores over collected data in order to make a decision. This means that machine learning becomes even more effective when you have more data.
Machine learning is often associated with smart homes, but this is not always the case. Machine learning is not limited to the devices you have at home. If you have a Facebook account, or if you use Google, chances are, you have experienced machine learning first hand. These two uses machine learning to help organize volumes of data it gets, and this helps both Google and Facebook to determine which advertisements are relevant and interesting to you. Machine learning also makes searches a whole lot faster.
You may not have realized it before but machine learning algorithms are used by a lot of companies, so much so that it is not an exaggeration to say that you encounter it daily. Some examples of real world machine learning algorithms you may have used are when you see recommendations for another product when you buy a certain good, or having voice recognition software (such as Siri) understand what you are saying even if you have an accent or if you pronounce words a little differently than the rest.
Machine learning takes its foundation from neural networks. Neural networks are built for learning and training. It uses several factors of importance to know what you want or need. Neural networks might need to be programmed or trained by humans first, so that they would learn. For instance, if you ask your voice assistant to call Sara, it knows to search your phone book for Sara’s number. If you have two or more Saras in your phonebook, it will ask you which one you mean. Over time, because of machine learning, it will no longer ask you which Sara you want to call and just goes ahead to dial.
In short, machine learning can train itself to be more accurate, without needing any human intervention. At first, you might have to “train” it, but it will learn as time goes by. Not only that, machine learning is also able to sort inputs and gives you accurate results even in real-time.
While we agree that machine learning is pretty awesome, you would realize that it does not involve any real intelligence. Algorithms are just a sequence of numbers, logic, and equations that do not see the need to improve or correct itself. It will just do what it is programmed to do, but with the accuracy of improving as time goes by.
When machine learning has gotten enough data, in such a way that it can interact with humans as if it were human itself, and even makes decisions on its own, then machine learning makes way for artificial intelligence.
In short, it would be impossible to have artificial intelligence without machine learning, and this is the reason why these two are often – mistakably – lumped together and confused with each other.
Imagine a machine that is able to identify and group together photos of dogs. An algorithm is programmed to look for certain characteristics that would tell it the photo is that of a dog, then you feed it with several photos of dogs. The machine will learn how to identify photos of dogs and it will eventually learn that what it is seeing are dogs, helping it identify Fido. That’s amazing, but it is not something that we would call intelligence by human standards. Even toddlers and babies can identify dogs when they see them.
If you put that algorithm in a system where it has access to cameras and speakers, allowing it to see the objects in front of it and then be able to respond to questions. And it does so in such a way that it mimics human behavior and intelligence. It is no longer just machine learning but artificial intelligence.
Types of Artificial Intelligence
There are two major types of artificial intelligence:
- general artificial intelligence and
- applied artificial intelligence.
Applied artificial intelligence
Applied artificial intelligence is more akin to machine learning, but with computers making decisions for itself. For instance, if you use LinkedIn Messaging, then you would know that it could give answers and replies to certain messages. The system uses predictive NLP, or natural language processing. What it does is that it’s looking at the message and looking for words that it understands, such as “new job”, “promotion”, or “new baby”; and then gives you reply suggestions based on these words. In order to do that, it would need to be trained using huge volumes of messages that allow it to learn the most common responses to certain messages and apply these to what it “reads”.
General artificial intelligence
General artificial intelligence is much broader than applied artificial intelligence. Rather than just relying on algorithm, general AI would need the machine to interpret, understand, and respond on its own. This could involve a variety of stimuli, as well as a number of tasks. In short, it would require the machine to think and act like a human. There are not too many examples of general artificial intelligence that we have today, but with applied artificial intelligence, you could probably go on and on with your list of examples.
Photo courtesy of Royce Milam (Flickr).