Science fiction books and movies routinely depict robots that can think, act, and even pass for human. It sounds impossible at first to build a machine that can think and act like people do; the human brain contains billions of neural connections, and we still don’t completely understand how it works. But artificial intelligence may be closer to reality than you’d think.
We’re still not able to produce a program or a robot that can make intuitive guesses, infer things about others’ states of mind, develop preferences or beliefs, or perform many of the functions that supposedly make human and animal brains unique. But what seems like an exclusively human function today can become possible for machines. It’s happened before.
Decades ago, chess was considered an activity only humans could master. Only humans, it was believed, had the cognitive ability to recognize patterns and predict an opponent’s next move. Today, you can buy an affordable computer chess game at Radio Shack that can beat a chess master.
The term “artificial intelligence” is usually used to refer to machines that can think like humans. So far, no such machines have ever been successfully developed. However, we’re making progress—and scientists have been successful at modeling different facets of human intelligence in software. Here’s an overview of the progress already being made in artificial intelligence.
Cognitive robotics. The discipline of cognitive robotics seeks to build computing software that can not only process information, but also exhibit signs of preferences and beliefs, goal setting, planning, revising plans, reasoning about its own mental state and drawing conclusions about the mental states of others, and collaboration in order to achieve complex goals—the way humans and animals regularly do. While nobody has yet built software capable of analyzing data this way in a real-world environment, some researchers have managed to create programs that can perform some of these functions using controlled data input from the lab.
Computational intelligence. Computational intelligence seeks to build machines that can recognize patterns and make predictions. Chess programs are examples of computational intelligence. Programs that use computational intelligence are being developed for industrial quality control; facial, voice, and handwriting recognition; analysis of scientific results; and even medical diagnosis as well.
Data mining. Data mining programs sort through large amounts of data, analyze it, and pick out information judged to be relevant. One common application of data mining is in business, where it is used to evaluate which customers are most likely to leave the company or respond to an offer. Businesses can then decide which customers to call or advertise to based on the information provided. Data mining programs are also used in financial analysis firms, law enforcement, and even search engines.
Artificial brains. There are several projects in the works to build an artificial brain. One example, Project Blue Brain, is a collaboration between IBM and the Brain and Mind Institute in Switzerland. The project seeks to create a computer simulation of the human brain down to the molecular level. It sounds like an impossible task—the human brain has billions of neurons—but they’ve already been able to map out a rat’s neocortal column.
Robotics. The brain is only one part of building an artificial being. The body is the other half of the puzzle, and putting together a robot that moves and acts like a human or animal is no easy task. Robots must be able to move across surfaces that change in texture and level; sense and navigate around obstacles; manipulate objects; and communicate with humans in a variety of ways by recognizing voice, tone, and even facial expressions. Some robots have been built that can produce a variety of facial expressions. In addition, commercial “pet” robots have been developed that simulate personality and emotions.
We’re still not able to build robots that can pass for human. But progress in robotics and in artificial intelligence is making steady gains, and companies, governments, and even individuals use aspects of it every day. It’s not unreasonable to project that artificial intelligence will continue to develop and improve until we can make a reasonable simulation of the human brain. Within a few more decades or even the next century, the science fiction stories that seem unlikely today may become everyday realities.
Artificial Intelligence:How Close Are We?