What is Artificial Intelligence (AI)?
The concept of artificial intelligence was created in the 50’s, when Alan Turing created a test to determine whether a computer had real intelligence. Soon after, in 1956, the Dartmouth Workshop took place and was the first AI conference. This fired the official starting gun for this new field of science. In this conference, speculation was raised about how far learning (in all its aspects) could be described in sufficient detail in order to be reproduced by a computer.
The fundamental idea that artificial intelligence is based upon creating a computer that is capable of solving a complex problem as a human would. Occasionally, these “complex” problems are not so complex for a human. For example, it is very easy for a person to:
- Identify a car in a photo
- Read a blurry text, or one that is missing a letter
- Identify a sound
- Prioritize tasks
- Drive a car
- Play a game and win
- Doing something creative such as writing a poem or drawing
In order to emulate the process, a computer needs to bring together its processing speeds, data storage capabilities and calculating power and others in order to imitate the reasoning of a human mind. Therefore, it must be able to:
- Take in information from its surroundings: Perception. As humans we communicate with our environment by using our senses. Nowadays there are a multitude of sensors that can carry out this function, collecting information from it’s environemnt and sending it to the computer for processing. They can even overtake human senses, since they aren’t restricted to our biological limits.
- Understand natural language: Natural Language Processing. The ability to interpret spoken and written language, to understand a phrase’s meaning and distinct accents. This is a difficult task, since the meaning of a phrase can vary with context.
- Represent knowledge. An AI capable of perceiving people, objects, concepts, words and mathematical symbols needs to be able to represent this knowledge in its “artificial brain”.
- It also must be capable of reasoning. It must connect all this knowledge, data and concepts in order to solve problems using logic. For example, a chess machine detecs the movements of the pieces on the board and, by applying the rules of chess to the data it has collected, decides on the best play.
- Being able to plan and move. In order to truly appear human, thinking like one isn’t enough. An AI must also be able to move in a 3D world, choosing the best route. This is what self driving cars already do, and it’s important that they do it well because errors in this case cost lives.
What factors have helped AI to develop?
One of the factors that has contributed most to AI’s advances, as well as the investment in R+D, has been Moore’s Law. In 1965, Gordon Moore predicted the continued increase of the complexity of integrated circuits (measured by the number of transistors on a computer chip), while reducing their cost. This allows the then-growing industry of semiconductores to create the microprocessor (in 1971) and other integrated circuits that were initially applied to computers but are now found in any device (phones, TVs, cars etc) and even in living things (such as animal identification chips). Thanks to thins, applications of artificial intelligence from a part of our daily lives.
Another factor that has helped AI to develop has been Big Data technologies. In 2012, Google showed that it was capable of identifying a cat in a photo with 75% accuracy. In order to achieve this, they used neural networks which were trained using 10 million YouTube videos. This would not have been possible without Big Data.
What types of AI are there?
Essentially, there are two types of artificial intelligence. Those known as “narrow/weak AI” are characterized by being “specialized” in a concrete task (such as winning a game). Deep Blue, created by IBM, won a game of chess in 1996 against the grand master Gary Kasparov. In 2016, DeepMind’s AlphaGo, created by Google, beat the south korean Lee Sedol in a game of Go. Digital assistants such as Siri and Cortana are also examples of this type of AI. They can show us weather predictions, or recommend an alternative way of getting to work, but they cannot read our messages and delete the unimportant ones. They can’t go beyond what they were initially designed to do.
“Strong AI” takes us into the world of science fiction. An excellent example is Samantha, the personal assistant of Theodore Twombly in the movie Her. Samantha is the perfect personal assistante, because she can learn new things and modify her base code. She can organize your emails and meetings, beat you at chess and write your shopping list. She is intelligent, empathetic and adaptive.
We have seen that the concept of AI is a wide one that touches on many distinct areas. In reality, we could define is as an ecosystem, where we are able to find technologies such as NLP Text Mining, Deep Learning, Predictive and Prescriptive Analysis, Machine Learning, recommendation systems and more.
All of these technologies are characterized by having something in common: they generate data with which you can uncover value and understanding. Because of this it is said that artificial intelligence in in the center of all these solutions, in the point where everything converges.