Artificial intelligence is here, but we don’t even know what it can do yet.
By Emilie Oblivión.
The artificial intelligence boom has arrived, not with the bang of omnipotent robots doing our bidding, but with the whisper of computer processors, aiding us more and more in our day-to-day lives until we can’t imagine a world without them. Image recognition, product recommendation, and satellite navigation are just the beginning of this technological revolution.
The current definition of artificial intelligence (A.I.) is broad, and what is considered an intelligent machine is up for debate. Classification of A.I. is based the two domains in which it can operate: closed and open. An open domain bot, whether in the form of software or an autonomous unit, can function in any field. There is no restriction on what the bot can process in terms of their general environment. Open domain includes sci-fi machines like the droids in Star Wars. They can apply their intelligence to diverse tasks as easily as a human can, for example, jumping from writing book reviews to predicting weather and stock fluctuations. This may be the end goal of A.I., but with current technology, it is almost impossible to achieve because it requires the bot to process such large sets of data.
Closed domain A.I., on the other hand, may not seem like the robots of science fiction, but it is the way artificial intelligence has already become an integral part of our society, and it is increasingly becoming indissociably so. These A.I.s have expertise on a limited subject, for example, automatically calculating the fastest route across town or finding and comparing flight prices. Google Maps, however, cannot simply begin piloting a driverless car. It is only intelligent across a limited number of tasks.
Although A.I. has been a concept in development since the 1950s, it has only recently made serious progress. This is mainly due to the breakthrough that machines can now learn by themselves instead of being programmed by hand-typed code. The previous method not only made the coder’s job tedious, but it also limited the capabilities of machines to exactly what coders told them to do. Where as previously programs could only …
To read complete articles from the Winter 2017 issue, subscribe to our eReader edition.