The press has reported dire warnings coming from some prominent scientists and technologists about the potential catastrophes the development of artificial intelligence (AI) may bring. Phrases like summoning the demon and end of the human race are being bandied about. Really? Don't you think this is a bit over the top?
As embedded systems developers we are used to calling our designs “smart,” but the truth is that they are nothing more than machines following an algorithmic program. Even the most powerful systems we have created only produce results that simulate intelligent behavior. The program “Eugene,” which is said to have passed the Turing Test, simply responds to typed user questions. It is not capable of independently composing a question much less planning the takeover of mankind.
The famous Watson of Jeopardy fame could answer questions quickly but did so without understanding. It simply did keyword spotting and calculated a relationship tree based on those words, choosing the response that had the greatest correlations. In the televised competition it would sometimes repeat a wrong answer already given by a human opponent, a real bonehead move. Further, the program only played Jeopardy; it couldn't learn to do anything else.
To read the rest of the article, visit EBN sister site EDN.