Artificial Intelligence: A Brief Rundown

    Machine learning is a critical part of Artificial Intelligence as we know it. Much of modern AI schemes are developed to statistically gather information on a matter and build on the information received. There are many methods, and some more effective for certain tasks then others. To some, AI seems like an exotic concept, far on the horizon. When I told a friend I was taking a class on AI systems, his first question was "Are you building Skynet?" AI is a reoccurring theme in popular science fiction literature and film, and as such has developed a bit of misconception. Scientists aren't trying to end the world with a super computer run wild, and really aren't even close. However, there are important developments in the field that have allowed for significant improvements in management of big data and regulatory systems through concepts such as machine learning, or Weak AI.

    The field of AI research has been around for a long time, having been founded in 1956. The holy grail of AI was then, and is still now, to develop a system that could emulate the human brain, and be capable of reasoning on the same level as a human being. Researcher's initial, and in hindsight optimistic, predictions stated that this machine should exist in less then a generation. The prospect of machines being able to do all of our most dangerous or menial jobs was enchanting, and garnered massive government funding for the time. Researchers had completely underestimated the difficulty of their task, however, and by 1973 funding for the projects had stopped almost completely. Research continued, despite the setbacks of what was known as the AI Winter.
Today, you could say we have come out of the winter and into the spring of AI research. Problems that were deemed unsolvable in the heyday of AI in the 60's have been solved, and applied to various technologies. Milestones include autonomous cars, chess programs, and more recently IBM's Watson.

    AI Research has fragmented into a variety of sub fields, focusing on individual problems and concepts. Machine learning is one of these, and has shown to be the most practical field. It is the study of computer algorithms for learning to do things. Data is observed, and the algorithm tries to learn based off this data to do better in a task in the future. The data may be fed manually, but the entire decision making process is emphasized as automatic. There are limits to what problems can be 'learned' in this manner, and the easiest tends be classification problems. These include character recognition, facial recognition, language understanding, spam filtering, and fraud detection.
 
There are some philosophical issues dealing with AI development, and our society as a whole needs to face these as we approach the holy grail. They are tough questions, and I will readily say I'm not qualified to tackle them! I just know if we make some Strong AI with a godly intellect, I would ask that it be restricted access from our nuclear stockpile.







Sources:
http://www.i-programmer.info/babbages-bag/297-artificial-intelligence.html
http://www.cs.princeton.edu/courses/archive/spr08/cos511/scribe_notes/0204.pdf
http://library.thinkquest.org/2705/history.html
http://aitopics.org/misc/brief-history