Ready. What did you imagine? Let me guess: in your head, Artificial Intelligence is a robot shaped like some weird AirPod, all surrounded by numbers and with a blue background. Am I right or not? In your defence, that’s normal. Sometimes it’s so difficult for us to imagine what Artificial Intelligence is like, and understand how it works, that we resort to fiction, to what popular culture has implanted in us as an image of it.

We think of Ultron, that machine created for good but that, in the end, turned out to be the villain in the movie (which is why it’s called The Age of Ultron); or Samantha from Her, that device created to ease loneliness that ended up stealing the heart of the poor protagonist, who felt completely alone. But, in reality, nothing could be further from the truth. In fact, AI has never been, nor is it a thing from (only) films. AI is a tool we use every day.

We can find it wherever we look: on Alexa and Siri devices, on your trusted Maps application, in video games, and even in public bodies’ scholarship and grant administration programmes.

If AI is so popular, fascinating, and common, then how come we get so confused when we try to identify it? Below we’ll try to explain how all this technology works, and then we’ll give you an answer to the million-dollar question: what is AI and, what is it not?

THAT AI THINGY

AI is an area that’s constantly evolving. Since its inception in the 1950s (yes, believe it or not) its definition and objectives have been changing, just like the technology that has evolved in relation to it. According to Simon Allardice, a professor of Software Development, the goalpost is always moving in AI.

Even so, the definition given by John McCarthy, a computer scientist and one of the fathers of Artificial Intelligence, is still considered valid today. This definition was the first to exist, and it came out of a conference (Dartmouth, 1956) in which everyone in attendance (including McCarthy) became pioneers in this area. It says that Artificial Intelligence is the following:

“The science and ingenuity of making intelligent machines, especially intelligent computer programs”

(John McCarthy, 1956).

Now, knowing broadly what AI is based on, let’s get to the point…

TO ANSWER THE BIG QUESTION… WHAT IS AI?

• AI can be strong and weak. For now, AI is just weak. When we say “weak AI”, we are referring to Artificial Intelligence that focuses on more specific tasks about which it can learn. It is pre-programmed by a person, in a specific environment, trying to simulate predetermined functions of human behaviour. To put it another way, it’s everyday AI, programmed by people to serve people, imitating them. We can cite the example of a bot that plays chess and learns from each opponent it faces, or other more common things, like automatic translators, the system that recommends songs on Spotify, driverless cars, and our friends Siri and Alexa. Even so, scientists are trying to develop a type of Artificial Intelligence that is not only capable of emulating what people do, but also capable of learning. This is what we call “strong AI”, and we could give as an example Samantha from Her, or any other form of Artificial Intelligence from films we can think of. We gave the example of a film here because this type of AI is still a very experimental concept. As if it were a child learning through life experiences, the challenge is to try to make the machine in question learn things with all the input it receives – not just copy and paste what people tell it. That is, it must be autonomous, intelligent, and multifunctional.
• AI is a glutton for data. If you don’t have data, you don’t have AI… that’s for sure. As we’ve said, we haven’t yet found the sweet spot to be able to create Artificial Intelligence with its own mind and capable of thinking for itself. Thus, all AI needs to gobble up data in order to learn from it and, subsequently, from human beings. This data can be structured (data organised in a way that facilitates its processing) or unstructured (data in the form of video, audio, or any other way that requires specific processing in order to be managed) and, depending on the level of the data’s structuring, the AI’s algorithm will use one learning technique or another to draw conclusions. In recent years, companies and organisations have been collecting data that has been very useful for the evolution of AI and Machine Learning, which we will talk about in a moment.
• AI is surely at the top of its class in math: algebra, calculus, statistics… this little brainiac’s good at it all! All these types of math are required to create even the most basic Artificial Intelligence. Algorithms are what breathe life into the beast, giving AI the ability to learn. Simplifying things, an algorithm is defined as a set of operations that allows a calculation to be made and the solution to a problem to be found. Algorithms are like the instructions that are given to the machine so that it can find the answer to something. These algorithms are then used in programming languages, which are tools that help to program Artificial Intelligence. From a developer’s point of view, there is no perfect tool for AI programming, but the most widely used are Python, C++, R, and Java. Traditionally, it was very expensive to program AI because the disks that had the requisite capacity were very costly. Nowadays, however, there are platforms like Microsoft Azure, AWS, and Google Cloud Platform that offer large storage spaces in the Cloud where one can pay as they go, only for what they use. There are also some free platforms, such as BigML, that make AI more accessible by anyone who wants to get on the bandwagon.
• AI is a modern-day fortune teller. In this day and age, we can anticipate almost any situation through Artificial Intelligence and the possibilities it gives us. We can know what kind of clothes to take on a holiday that is still a week away, we can invest our money in what we believe will be the best option, and even anticipate a problem with our own health that could save our lives. These are just some examples of what AI has achieved in recent years. And, although this has been a gigantic step, giving us stability and security, just like all the people in the world, sometimes the AI is wrong. The fact of the matter is that probability and reality are two totally different things, depending on many other factors than just statistics. Even so, AI has managed to use data and statistics to give us a certain “glimpse” into the future on many occasions. A very clear example of this is that of the platform known as Delta 8.7, which uses AI to detect patterns and apprehend people who engage in human trafficking.

SO, THEN… WHAT ISN’T AI?

• AI is not Machine Learning (ML). Or at least not all AI is Machine Learning. AI is a much broader term than the one just mentioned, as ML is only a branch thereof. To define it somehow, Machine Learning is a fundamental process by which AI acquires part of its intelligence by automatically learning from previous data without the need for prior programming. We could say that it’s AI’s school. ML uses huge amounts of data to come up with results or predictions based on a learning model. Therefore, we could say that all Machine Learning is Artificial Intelligence, but not all Artificial Intelligence has to be Machine Learning. This type of technology is what is applied for face detection, or to sort through spam.
• AI is not Deep Learning. Deep Learning, a concept that falls within ML, is another way in which machines can learn, but there is a difference with the method just mentioned. In Deep Learning, algorithms known as neural networks are used, which are not AI either. Although it would be cool to say that these neural networks emulate the human brain, for now that’s going too far. Let’s just say that they’re inspired by a tiny part of the human brain. A neural network is composed of several interconnected layers. Each “neuron” passes information to another, and another, and so on until data is gathered that helps the machine acquire new knowledge. This is the technology used by cars that are able to drive themselves.
• AI is not capable of conscious thought (for now). If these machines were edible, vegetarians would be allowed to eat them, as they’re still lifeless beings with no minds of their own, needing people to survive. The human factor continues to be one of the most important features of Artificial Intelligence. In fact, being an invention of humanity, AI tends to replicate the behaviour of people, which is a double-edged sword. This is what happened to SimSimi, a cute yellow chatbot designed to provide company that ended up becoming a monster that collected the darkest data from the Web. There are several initiatives to prevent this from happening and to ensure the proper use of AI. The “Partnership on AI” group, for example, has more than 100 members (including Apple, Microsoft, Amazon, Google, and Accenture) dedicated to creating and following certain guidelines so that everyone can enjoy safe and fair AI.
• AI is not a robot. It goes far beyond that. Robotics is the science that is responsible for the creation and maintenance of robots, and although some are programmed with Artificial Intelligence built in, not all have to be. Although there are robots like the Roomba, or Penny (the robot that serves you food in a restaurant) that combine robotics with Artificial Intelligence to make the daily life of their users more convenient, there are others that do not use AI. Examples include the Thermomix cooking device, the robotic arms used in factories, and even the vast majority of robots used in physical fights between each other, proof that not every robot has to learn in order to be functional. Sometimes just doing what they’ve been programmed to do is enough.

CRYSTAL CLEAR?

After reading this text, we hope that Shakespeare’s question of “to be or not to be” is more or less answered – at least as far as Artificial Intelligence is concerned. The original question has no answer; it is simply the author’s way of creating an existential crisis for everyone who reads Hamlet. And, although clear answers may have been given in this text, AI is an industry in which everything can suddenly change from one day to the next… so, who knows? Tomorrow all the points we have discussed may be totally different thanks to the appearance of a technology that has created an “Ultron” that thinks for itself. Thus, even if “to be or not to be” is clear, “what exactly will be” is a little more complicated.