After decades of a shadow existence in theoretical computer science, Artificial Intelligence has suddenly made an entrance into the world in the last few years. The human-like intelligence created within a computer is what most people think of when they hear the term Artificial Intelligence. But the truth is that so far, our efforts to create such an Artificial General Intelligence have been humbling. Many problems arise in training an entity with no body or concept of physical existence to act like a human. What is often overlooked is the cousin of these super-computers, known as machine learning AIs or algorithm-based AIs. They are what we call ‘smart’ nowadays, which means they are programs with the ability to make their own decisions, based on the data they are provided. These small, smart machines will probably have a much larger impact on humanity than anybody can imagine at this point.
The AIs of today are not conscious superintelligences, but extremely effective and well-trained digital worker drones. And they are everywhere.
Algorithms now decide what movies we like and what we want to buy. They can create a fake Tom Cruise on TikTok, train our children in games, give us customer support and write articles for The Guardian. They guide robots, search the internet for terrorists and analyze potential voter behavior. Machine learning AIs are getting better at making decisions every day. They are a combination of trained algorithms following a rigid and complex set of rules with access to databases and sensors to get the necessary data to make their decisions. They will soon drive our cars, fight our wars, and watch our homes. What most people only begin to realize is we already interact with them every day. Even on a simple mobile device, little machines are analyzing our browsing behavior, our social preferences, and personal interests, to make custom-made offerings to us.
AIs are trained, which means they learn. Learn what? Basically, they analyze human behavior. If we do A, how likely are we to do B?
So, the question that arises, simply put, is where this will lead?
Quite simply: A takeover. A very silent one.
Not one with weapons and killer robots, as Hollywood has imagined it. AIs mean no harm. They mean nothing. The world will be handed over to them by us, so why would they ever need to conquer it? They basically do what they are trained to do. We train them, for now. Their limits today are the limit of their creators to train them for certain purposes. We use them because they are fast, cheap, and effective.
That is only the beginning because somewhere on the horizon lurks the ultimate application of artificial intelligence: the ability to train other artificial intelligence systems. It is called Singularity when machines create machines. Once this stage is reached, their growth will be exponential, probably only limited by available computing resources. What will be the result of this new type of AI?
AIs are not smarter than us. Faster, yes: computers can run millions of operations in seconds. AIs that can train their own will lead to more and more sophisticated machines. They will be able to draw from millions of records of behavior and speech patterns and imitate human behavior to perfection, such that they will be impossible to distinguish from a human when interacting with us.
At this point, we will be embedded into a system of AIs that know every need we have and every preference and habit we develop, and they will be able to act upon it. They will design our surroundings to fit their insights. Computers will take over a large part of our daily decisions. What, then, will be our role in that society? Well, we will provide the data they need to fulfill their purpose. They will be in our houses, cars, fridges, mobile devices, games, and clothes. An omnipresent system watching over us.
It will probably be a comfortable life in which our fridges fill themselves and our cars get us wherever we want, our finances are managing themselves and they even do our taxes. The problem is, we will also be embedded in a system that will decide what we see and consume. We will make fewer and fewer decisions ourselves. In many ways, humanity will be reduced to data, processed in a system of highly effective and hardworking little machines that know everything about us.