Computers, AI, and Rise of the Machines
Computers overtaking the world is a common theme in science fiction. People seem to be fascinated by the idea that humans may someday be destroyed by their own creations. While there are scientists that genuinely feel this way (Stephen Hawking, for example), it is safe to say that such possibilities are far in the future.
The truth about artificial intelligence in modern practice is that it is non-existent. There are many news articles and tech companies that love to slap the term “AI” onto their software or products—and they get away with it because the general public doesn’t fully understand what AI really means. To be fair, the concept of classifying AI is a gray area, even among experts and engineers alike.
The broad definition of AI is “intelligence exhibited by machines” but what is intelligence? Some believe that intelligence involves self-awareness and an understanding as to why an individual takes action. Others believe intelligence requires an ability to be logical and solve a situation. Computers can solve many problems faster than people can, but are computers intelligent? If you ask in these terms, the answer is no. A computer is simply following instructions and, when computers run programs, they are not aware of their actions.
For example, a computer that beats a human at chess is not intelligent as it has no understanding of chess. In fact, the computer is not aware that it is even playing chess and is just executing many millions of if statements.
However, just because real AI is currently not in operation and computers are not truly intelligent does not mean that computers are incapable of human tasks. It also does not mean that algorithms and programming methods are becoming more like the human brain. Neural networks involve programming techniques that allow a computer program to adapt to inputs, learn from outputs, and become better at tasks.
A classic example of such computer brilliance is Google's “AlphaGo” which can beat players at the Chinese game Go. AlphaGo recently went on a spree, defeating Go masters anonymously. At the beginning of this year, it was revealed to the public, as well as to the masters it had defeated, that this mysterious Go player was an AI all along.
Originally, the problem with solving Go was the sheer number of moves that can be made (approximately 1 x 10^170, which is more than the total atoms in the universe). This means that a computer cannot use brute force for a winning strategy as a human will have the upper hand thanks to intuition. Therefore, the Google team took a different approach and developed a system that could almost be considered a form of synthetic intuition.
Instead of trying to calculate moves, the system essentially runs different tasks, such as identification of important moves, and found a balance between a neural network, a policy network, and a value network. This balancing is similar to how the human brain makes decisions, where different aspects of a situation are considered. For example, a person playing chess does not always think of every possible move and plan according to immediate consequences. That player could allow moves that may seem bizarre to a computer such as self-sacrifice which, in the short term, appear to make the player lose but in the long term could be vital for a victory. Other techniques include confusion whereby a player that is losing may make “crazy” moves that can confuse other players and keep their true strategy hidden.
Go: an ancient Chinese game. Image courtesy of Wikipedia
Puzzle Games Are the Key
Computers playing puzzle and board games such as Go and chess demonstrate how computers may not be far from ruling the world. Currently, the world as a whole is already littered with electronics ranging from IoT devices that track the temperature of a room to internet-enabled toasters. As this integration and demand for more intelligent systems increases, it won’t be long before computers are essentially in charge of everyday life.
It is at this point that computers could start making decisions for us on a global scale that we may not be aware of. Farms, for example, are increasingly becoming dependent on computers to track production rates and record vital information, including nutrition levels in the ground and pest control figures.
Imagine a computing system that takes data from weather probes, IoT temperature and humidity sensors, and farming data. Such a system could make determinations as to when it will rain and therefore be the ideal time to apply fertilizer or other needed chemicals. The result would be highly efficient farms that use neural networks to make all the key decisions without any human interaction at all.
Settlers of Catan is a board game that deals in the trading and amassing of resources.
Some scientists from Tilburg University have claimed to create an AI that can play Settlers of Catan in a competitive manner. For those who may not be aware, Settlers of Catan is a board game that involves players creating villages and cities while competitively trading resources. The real world is not so different from Settlers of Catan with countries across the world trying to strike trade deals and improve their economic conditions while growing towns and cities to increase the size and importance of their own country.
While all these decisions are currently made by people, an AI that understands this relatively simple board game implies that maybe there is a chance that computers could begin to make such decisions when fed information about real world scenarios. Leaders could then act upon the advice of a computer instead of experts. After all, if AlphaGo's taught us anything, it's that AI is capable of “playing the game” better. It's feasible that an AI could, in fact, orchestrate the gains made from trade deals better than a human expert could.
Computers that play games are not entirely intelligent and no system on earth can be considered as being truly intelligent as of yet. However, engineers and programmers alike are really not that far from producing a computer that could make real world decisions on behalf of people—and do a better job at it. Of course, depending on who you ask, this is how it starts.
First, computers start to learn about your central heating habits and tell you what temperature you want your home to be. Then, before you know it, they are ordering a pizza for you and choosing your toppings without even asking you. The worst of it is that you probably won't even be mad that it got it right.