Artificial intelligence (AI) started as an academic discipline in the mid-1950s. Since then, advances in technology have allowed us to have forms of artificial intelligence in our homes – on deep learning workstations, smartphones and televisions. But when it comes to the different types of AI, including machine learning and deep learning, how are they different from the technology that is used in the in-built chess game on your PC?
To start with, we should clarify what AI, machine learning and deep learning are. The first thing to note is that they’re all connected and all considered part of the same area of study. They’re also all capable of different things, depending on how complex they are.
Caption: Diagram showing the connection between artificial intelligence, machine learning and deep learning.
What is AI?
According to the dictionary, artificial intelligence is “The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”
AI is a computer that mimics human behaviour. They are currently only capable of performing the specific tasks for which they are programmed; scientists refer to this as Artificial Narrow Intelligence (ANI). Artificial General Intelligence (AGI) would be if AI were on par with a human, while Artificial Super Intelligence (ASI) would involve AI surpassing the intelligence of humans and becoming self-aware. Only ANI exists currently, and it is considered a weak AI compared with AGI and ASI, which are considered strong AI.
In general, it is accepted that weak AI cannot move past their coding or learn beyond their programming. Weak AI includes ones that complete very specific tasks, like winning a chess game.
What is machine learning?
Machine learning is a subset of artificial intelligence – this means that it is a more advanced version of AI. It is a series of techniques that enable computers to analyse data and deliver AI applications. The main difference between ANI and machine learning is that while the former imitates how humans behave, the latter imitates how they learn.
Imitating how humans learn means that if you give a computer with machine learning capabilities a set of data about financial transactions with the fraudulent ones indicated, it will then be able to identify the common markers of fraudulent financial transactions and identify them in the future as transactions take place.
What is deep learning?
Deep learning is a subset of machine learning. This simulates how the human brain works and learns, using an artificial neural network, which imitates the neurons in the human brain. This allows computers to solve more complex problems, such as recognising different objects when they change position. This could include being shown a picture of a horse in a field and then recognising a horse again if the horse is on a chair. The AI involved in autonomous cars and speech recognition is developed with deep learning.
Artificial Neural Networks
Deep learning is made possible by the existence of artificial neural networks. These imitate the neurons in the human brain to a much greater extent than the connections that exist in machine learning computers. Deep learning neural networks consist of at least three hidden layers between the input and output layers.
Caption: Image showing a simple artificial neural network
What are the differences between AI and machine learning and deep learning?
The main difference between these three types of artificial intelligence is how advanced they are and how well they can learn. Deep learning AI is the most advanced and is able to learn beyond its programming due to its artificial neural networks.
Are there any other differences?
1. Human involvement
Machine learning is less advanced than deep learning and requires far more human intervention to learn, unlike deep learning, where the machine can identify features and classify them without help. Some forms of AI are even less advanced than machine learning and cannot function outside of their programmed purpose.
Deep learning requires complex machines with powerful components – graphics processing units, primarily. Machine learning can be run on normal hardware due to it being less complex than deep learning. Simple AI programs are often installed on our computers as an in-built opponent for games like chess.
Deep learning systems take the most time to set up, but when they’re going you’ll save that time through how quickly they can generate results. Machine learning is often quicker to set up, but the results are often limited by the power of the machine.
4. Approach to problem-solving
Deep learning examines large amounts of unstructured data through its artificial neural network. This means it solves problems in a different way to machine learning which uses traditional algorithms to solve problems.
Machine learning is used for simple tasks like identifying spam emails, fraudulent transactions or to show films and tv shows you might be interested in on streaming services. Deep learning is used in self-driving cars, robots that can help in surgery or even to predict the risk of disease in humans in the future.
The main difference between AI, machine learning and deep learning is how much they can do without human intervention. Deep learning can do the most, AI the least, while machine learning is more in the middle. Other differences include their application, how resource-intensive they are and how long it can take to set them up.
It is important to remember that while deep learning and machine learning are more advanced than artificial intelligence as a whole, they are still types of weak AI. While we may not have AI that is on par with, or even superior to, humans yet, technology is advancing and we may see strong AI in the future. However, strong AI is not without its risks, as scientists like Stephen Hawking warned.
Author bio: Rachel Gowland works at digital marketing agency, Tillison Consulting. She’s a passionate gamer and avid reader who loves to travel, using her knowledge of foreign languages to connect with people around the world.