IBM’s ‘Watson’ artificial intelligence software is currently at the forefront of cognitive computing. It was first thrown into the spotlight of the public eye when it beat a human being in a jeopardy episode in 2011 but it has come a long way since then. Up until now, most artificial intelligence software systems have been primarily based on mathematical equations, rules, and logic; often following a very rigid decision making system. However, with the amount of information for these softwares to sift through constantly expanding, these systems are becoming more and more obsolete. Rather than this approach, Watson is dependent upon natural language which is governed by things like rules of grammar, context, and culture.
Watson reads and understands the information the same way that people do. It breaks down the sentences via grammar, relations, and structure which makes it able to actually understand context unlike simple speech recognition software like a google search engine that simply looks for key word matches and synonyms.
Due to this, unlike conventional computing systems which can only comprehend information if it’s neatly organized and well-structured, Watson can assess unstructured information the same way that humans do. Some examples of unstructured data are literature, articles, research reports, blogs, posts, tweets and pretty much any other types of data that is created by humans for other humans to consume. When humans want to understand something, there are four steps to which they attempt to do so:
- Observation of information and evidence
- Interpretation of what has been presented to them in order to form hypotheses
- Evaluation of which hypotheses are right or wrong
- Deciding which option seems best
Because Watson mirrors a lot of these same cognitive abilities that human beings possess, it approaches problems the same way that humans do but at a much faster speed and can assess information on a much larger scale.
In the initial stages of preparing Watson to become literate in a field, experts manually select the relevant information that is uploaded into the corpus of information. As time progresses, Watson’s corpus is constantly updated and improved in order to make sure that Watson’s information continues to be accurate. Once Watson begins operating in a particular field, it learns the language, dialect, jargon, and thought process of the field. This is done by giving it Question & Answer pairs that teach it how to find patterns of linguistics within the information in order to find the answers to questions on it’s own as well as identify new insights or patterns in the available information. Once identifying certain parts of speech within a sentence, Watson generates hypotheses which are then compared with the data within the corpus to either support or refute the hypotheses. Watson can handle even the most complex of fields such as law, medicine, or culinary arts. Currently, Watson is discovering patterns and offering answers to questions we never even knew existed faster than any person or group of people ever could.
Watson constantly adapts and changes the same way that human beings do. It gets smarter with age as well, just like human beings. It learns from interactions with human beings and from it’s own successes and failures just like us.
In a recent study at the University of North Carolina School of Medicine, Watson was given the task of analyzing the cases of 1,000 cancer diagnoses. In 99% of the cases, Watson recommended the treatment plans that matched the actual suggestions from real oncologists. Watson also found treatment options that the human doctors missed in 30% of the cases. Due to the fact that it can analyze thousands of documents in mere minutes, it allows the software to read all of the clinical trials or research papers that a human may not have the time to read.
Although about 2/3 of Watson’s sales are coming from the medical field, this technology has shown to be extremely flexible with a multitude of uses. Grammy-award winning producer “Alex da Kid” recently used Watson’s “Beat” algorithm to create his first solo EP. It has also created movie trailers all on it’s own. General Motors will be implementing the software for their newest OnStar system, OnStar Go. IBM is currently in the process of developing the software to analyze, identify, and hopefully prevent cybersecurity threats. Beginning next year, IBM will release a version of Watson that will assist teachers in creating lesson plans. It can also assist traders with “up to the minute” financial, economic, product, and client data for more personalized recommendations. These are just a few of the many ways this technology has been used.
A few years ago, I would have said that technology like this was not going to be accessible to the general public in my lifetime but I was wrong. It’s official, the intellectual capability of technology has now exceeded that of mankind. The future possibilities of this technology are endless.