Contraptions That Think

The human mind is complex machinery that is only now being understood. Not only does it make inferential conclusions and make judgments, it does so either instinctively or on the fly with near blinding speeds, as when facing emergency situations. Not only that. The human mind can adapt to its environment and more importantly, learn.

Understanding the human mind is the first step to duplicating its workings on what is increasingly becoming the fashionable tool for the 21st century office and factory – thinking machines.

But no doubt, it is one hard act to follow.

Engineers might be able to mimic some of the functions of the human brain, like enabling a computer do logical thought processes. They have certainly gone beyond that, like processing gazillion data to organize as a database that no human can do. But that’s just one, putting all that to fit into a box near the size of the human skull remains a daunting challenge.

Rules and More Rules

Computers are logic driven engines. Yes, to a large extent they mimic the brain when making mathematic and logical deductive ad inductive reasoning. Much of the engineering effort to duplicate cranial activity hinges on rule-based logic using binary data to store and process information according to a set of instructions called application programs or software.

Artificial Intelligence

Spielberg may have gone overboard with his A.I. movie that has human-like cyborgs for robots, but that is basically the objective of robotics. We’re still decades from being there. For sure, there are now super computers in many research and think tank companies that employ some level of artificial intelligence that have been known to make decisions and recommendations based on terabytes of data processed within minutes. These super computers do form the brain of any robotic contraction approaching human qualities.

Artificial intelligence programming is still rule-based but it has an integral database engine that can process large amounts of information into multiple parallel computers. They are definitely not your typical corporate mainframes, but a step higher.

Processing vast data, super computers use special application called expert systems that can mimic human intelligence efficiently in specific disciplines. Legal and medical expert systems processing terabytes of data are known to provide recommendations and diagnosis on the fly.

Neural Networks that can also Learn

Another way to artificial intelligence is the use of neural networks. Modeled after the human brain, they make use of programs that can handle ambiguities left out from rule-based systems. Human thought is not always based on black or white stimulus, but also allows options for what they call "grey areas." Unlike rule-based programs, a neural network can make decisions that are less predicable given a set of what if conditions. Their responses can lie in between using only the most probable answers.

Hence, we have what is called "fuzzy logic" computers which are sometime applied in home appliances. And neural networks can learn through exposure to various historical inputs and outputs generated from those inputs. Over time, they can be trained so that responses can be deduced from a set of rules of accepted output decisions that worked previously.