Have you ever asked Siri to schedule a meeting for you in your calendar? Are you still inundated with spam, or does your email provider self filter? Did you speak to a human the last time you made a call to your internet provider’s technical support service, or was the problem handled entirely by a machine?
It seems that when we hear the words “Artificial Intelligence”, the image that comes to mind for many people is that of humanoid robots who think for themselves, perform the work we actually do, or who will end up rebelling against us (thank you Hollywood for fuelling this image). But the truth is, the advancement of artificial intelligence is not a sign of some impending dystopian future, but is already, very much a part of our daily lives, even though we hardly notice it.
There are many definitions of Artificial Intelligence, but one of the most accurate describes it as: “[a response] to the theory and development of computer systems capable of performing tasks that normally require human capabilities, such as visual perception, language recognition, decision making and language translation.” For example, Siri is a language recognition system, mail filters are a method of decision making, and customer service bots are a mixture of both, and they all fall under the umbrella of Artificial Intelligence.
These applications fall into the category known as Weak Artificial Intelligence, which is based on supervised programming. This type of AI (still being developed as we speak) uses computer systems to perform repetitive tasks and resolve associated conflicts. In this way, a computer can learn, after viewing thousands of images of an object, to recognise that object in a new, unseen image.
On the other hand, and getting closer to the cinematic perspective of AI, we encounter Strong Artificial Intelligence; AI that is conscious and sentient. At this point, we don’t have any real examples, only our own imagination.
It is very easy to believe that Artificial Intelligence is only a small area of science and robotics, or that it is relatively new, but its foundations are based on a multitude of disciplines, including philosophy, mathematics, economics, neuroscience, psychology, control theory and, of course, computer engineering and cybernetics. All of these have influenced, to varying degrees, what has historically been understood as Artificial Intelligence.
Although the first mention of the term was in 1956, Alan Turing is credited for making early progress in this area. In fact, the concept underlying the theory of Strong Artificial Intelligence is that a computer is capable of passing the Turing test; namely, that a machine will pass the test when a human does not know how to distinguish his own answers from those that have been answered by a machine, that is, when the machine acts as a human being would.
Artificial Intelligence as a discipline has gone through periods of despondency, during which progress was stagnant and the idea of practical and valuable AI seemed unachievable. However, over the last decade, research efforts and technological advances, among many other things, have made it possible for a computer to pass the Turing test, or even to beat the world champion Go, the millennial strategy game.
This evidence may seem anecdotal, but it helps us to realise that Artificial Intelligence is advancing in leaps and bounds, with one objective in mind: to make our lives easier. Personal assistants are already a reality, but there are countless other applications under consideration, in many different sectors such as finance and medicine, which will bring about a real revolution.
Imagine the effect AI capable of diagnosing disease could have, or one capable of identifying when something falls off the production belt and stopping the machine, or one capable of issuing court rulings based on previous precedent, or, more immediately, a self-driving car.
What is being done to achieve this? Two words; machine learning. Machines are able to learn how to perform tasks from algorithms that compare huge amounts of repeated data. The two main elements of this process are Machine Learning and Deep Learning.
- Machine Learning is the development of algorithms whereby machines can learn from the data and subsequently make predictions or suggestions. Thanks to Machine Learning, computers are able to predict traffic, determine the best time to publish on social networks, or even identify which of our followers are most likely to buy our product.
- Deep Learning is a branch of Machine Learning that goes one step further: in this case, instead of giving the computer a series of patterns so that it can perform a task, it is taught a model so that it can extract examples and over time, learn how to correct deviations in the model, eventually becoming extremely precise.
It is in the sphere of Deep Learning that research is advancing most. The most influential technology companies are focusing their efforts within this area of AI. Google has created the platform Google AI, a website that serves as a meeting point for all those interested in AI. With tools such as the Google Cloud Platform, which provides Machine Learning services for companies, or TensorFlow, an open source research library, the giant has invested heavily in this technology in order to generate innovative ideas and integrate them into their products. Amazon has done the same with AWS, and Elon Musk is the man behind Open AI, a non-profit AI research company.
The efforts of these companies have led to improvements in many fields. At the beginning of this article we mentioned intelligent personal assistants such as Siri, Alexa or Cortana, as well as chatbots used for customer service. Even sectors traditionally less digitally focused, such as agriculture and logistics, have given in to AI. In the case of agriculture, there are layers of AI that predict the best time to sow, harvest and which products to use. In the case of logistics, autonomous driving and inventory management will improve delivery times, thus increasing productivity.
It’s not only businesses that will benefit from these improvements; on an individual level AI also promises to prove a turning point in society. In medicine, technology is being developed so that machines are capable of improving pre-diagnosis and tracking treatments.
Like any technological revolution, AI has been met with misunderstanding and fear. A great deal of news reports and articles have been dedicated to the potential job loss that will result from the implementation of many of these applications.
In 1942 Schumpeter made German sociologist Werner Sombart’s concept of Creative Destruction popular. This concept assumes that every time there is an innovation, new business models are generated to replace those that have since become obsolete. Redundancy following innovation has happened before, and it will happen again. In addition, new jobs are created based on the needs generated by new technologies.
Kevin Kelly said earlier this year in a TED Talk that any form of creative destruction involves doubts, but it always ends up advancing society. The same goes for the AI revolution. Kelly proposes a hand-in-hand model of collaboration between humans and machines: the perfect diagnosis will result not just from a computer, but a combination of a computer and a doctor.
Job loss is not the only shadow of doubt that hovers over the AI evolution. On a social level, questions have been raised about the privacy of data managed through AI; are the public prepared to live with this technology? Most critics argue that something inherent to mankind is consciousness, and that it is something which machines cannot take into consideration, thus eliminating emotional intelligence from the balance.
However, these doubts seem less pressing if we consider that they are more often related to the theory of Strong Artificial Intelligence. In order to make further progress, researchers face more technical obstacles The first is that the processing capacity of neural networks must be increased in order to enable them to carry out operations with the necessary speed and agility; the second is using huge amounts of data to extract models, and how this can be optimised.
It is clear that we have only just found our feet and are taking our first steps in the field of Artificial Intelligence; right now we cannot say what society will be like in twenty years. However, one thing is clear, no matter how much doubt we have, we can always turn to Siri for answers.