The human brain is capable of tasks with which the most complex computers have trouble. Comprehending language, a task that children can begin to develop by their second or third birthday, has only just reached a consistently usable level. Even with years of research, IBM's Watson, which competed against two all-time Jeopardy! contestants, was unable to detect many of the nuances of human communication. Puns, context clues, and more are critical parts of communication, but given several of Watson's "confused" answers, they were clearly misinterpreted.1 From this, an important question becomes apparent, "What makes a human able to understand this type of question (or problem), and how do we make computers capable of doing the same?" This question is not only applicable to language recognition, but also chess, packing a bag, finding an efficient route, and many other daily problems. These are all skills that we learn, practice, and than improve upon. This general concept can be extended to computers for computationally difficult and expensive problems like these. Teaching computers to learn and to apply previous results, like humans do, can be an effective model for one of their main functions, problem solving. This is the foundation of the field now known as Machine learning.
Nichols, Nick A.
"The Learning Process: Inspiration for Computer Scientists,"
The Intellectual Standard:
1, Article 5.
Available at: http://digitalcommons.iwu.edu/tis/vol1/iss1/5