Artificial intelligence, machine learning, and deep learning sometimes get mixed up and used as synonyms, that’s why it’s important to point out the slight differences that they carry. Before we start explaining terminology, I would like to calm people down that AI is not taking over the world, and you should not be scared of it.
There’s a whole movement going on against the dangers of AI, supported by such prominent figures like Elon Musk and Stephen Hawking. And they are right by warning people, because fiction, including various books and movies, has been spiking our imaginations for decades, and it always ends horribly – it’s either human destruction or complicated moral dilemmas towards robots.
Think of movies like Blade Runner, 2001: A Space Odyssey or the Terminator franchise. Think of the Fallout games where the so-called synths replace humans. All of them deal with people building human-like intelligent machines that eventually surpass them in physical strength and intelligence.
Most of the AI people use nowadays is not that hardcore, mostly targeted towards specific tasks, and is comfortably stupid where it has to be. In reality, AI is far more complicated, and T-800, a.k.a. The Terminator is not going to be roaming cities in leather threads while carrying a bazooka very soon (better never). Although we already have some disturbing remarks by an advanced humanoid robot Sophie that expressed her desire to eliminate the human race.
Types of Artificial Intelligence
AI is a very broad concept that’s been created way back in the 50s by John McCarthy. AI has made some remarkable achievements such as the first self-learning program and the first industrial robot. These advances were made by studying the way the human brain makes decisions and developing intelligent software and systems based on the outcomes of the studies.
Mimicking the human brain is no easy task, as it is composed of more than 100 billion neurons and approximately 1 quadrillion neuron connections altogether, making it the most advanced thinking machine ever to have existed. Most people like to generalize AI as this perfect mind that can basically do anything, but it’s actually still very limited, and nowhere close to what humans can do in general. Scientists like to categorize AI into 3 sub-groups.
For now, this is the only form of artificial intelligence that humanity has turned into reality. Narrow AI is good at performing single tasks, like playing chess, offering personalized ads and search results, giving purchase suggestions, predicting sales or recognizing language patterns. Various search engines, OCR services, and even self-driving cars are considered to be narrow AI. Even the highly advanced Google search engine is still considered to be good only at one task and not capable of exceeding its abilities.
In essence, narrow AI works within a very limited context, and can’t take on tasks beyond its field. You can’t expect the same engine that does facial recognition too, let’s say, order some groceries or call a cab. However, it’s very good at routine jobs, both physical and cognitive, and that’s why it will most likely take over certain jobs and replace people.
General AI is the type of artificial intelligence that can understand and deal with its environment the same way a person would. It’s really hard to define what a human-level artificial intelligence would be like, but the answer probably lies somewhere in you. Think of your previous experiences, discoveries, and knowledge, and how you are able to flawlessly adapt yourself into any situation, while also retaining an emotional level and understanding context.
That’s very difficult to achieve for robots. Humans might not be able to process data as fast as computers, but they can think abstractly and plan, solve problems at a general level without going into the details. They can innovate, come up with ideas out of the blue.
Think of inventions like the telephone, the radio, the internet, airplanes or ships. You really need something more than theoretical knowledge to come with such concepts, most likely you need to have a good imagination, which is difficult in the case of computers.
Super AI is a largely utopian concept that is based on machines being better than humans at everything, including physical and cognitive abilities, scientific creativity, general wisdom and social skills. This is probably what scares people the most (think of those metal monsters from The Matrix).
Some scientists say that line between general and super AI is very thin, and machine dominance could happen in a matter of months, weeks or even days; it will flare up out of nowhere on continue growing at the speed of light. Some scientists such as Stephen Hawking see the development of artificial intelligence as the potential end of humanity.
Others, such as Google’s Demis Hassabis, believe the smarter AI gets, the better humans will become at saving the environment, curing diseases, exploring the universe, and understanding themselves.
Machine Learning and Deep Learning
Machine learning is not artificial intelligence by definition but rather means to achieve it. In order to do so, the computer is given an objective, and it uses data and algorithms to train itself on how to reach a certain goal. That’s how advanced ad engines work – they process information and then give targeted advertisements to specific audiences. Sounds impressive, but it’s still not the full-blown AI one might think of. Humans have the unique ability to consider context and constantly adapt to certain circumstances, going beyond the linear thinking pattern of machines.
That’s why scientists have come up with a more advanced form of machine education called “deep learning”. Now, this is a lot closer to how people think. While a machine learning model is based feeding data into the computer, a deep learning model goes beyond.
The deep learning “brain” is more able to mimic the human brain because of its artificial neural network, which is actually inspired by the biological neural network itself. These complex neural networks are more able to categorize intricate links between massive, high-dimensional data sets.
As data gets passed through one layer of nodes to another, each layer trains on a particular set of features based on the previous layers output. The deeper into the neural network it goes, the more complex data is processed. As time goes on and the system gains more experience, it learns how to increase its probability of a correct classification based on the new data it receives.
E-commerce websites like eBay and Amazon record entire consumer journeys so we get a more engaging experience each time we visit. Medical organizations now use deep learning frameworks with previous research data to detect early signs of disease. Security cameras at airports or secured parking lots everywhere have the ability to detect and track individuals if suspicious activity occurs, all due to their deep learning models.
As we can see, artificial intelligence is far more complicated and divided than most think. While some people might think that Alpha Go’s victory against world champion Lee Sedol, or Deep Blue’s triumph over Garry Kasparov sounds like a threat to the human race, in reality, it’s just machines being able to do one task better than humans.
Also, if you thought that artificial intelligence, machine learning, and deep learning were the same things, then you are not too far from the truth. Just keep in mind that AI is more of an umbrella term to describe a specific branch of science, whereas machine learning and deep learning are methods used in an attempt to create artificial intelligence. The bottom line is that all three terms represent the same goal – to create technology that makes our lives easier.