Since the earliest forms of science fiction, authors and futurists have dreamed of robots with cognitive abilities matching those of humans. The interactions of these machines with the world of people would come naturally for both sides. Conversations with robots of the future are often imagined as indistinguishable from normal human-human discussions, since the machines recognize the subtleties of language and speech.
Unfortunately, the robots of science fiction haven’t materialized in the real world. Artificial intelligence can’t recreate some of the simplest human brain functions. And although the power of computers gives interactive robots far more computational efficiency than the human brain, it struggles to simulate the behavior and language of people.
IBM, known for developing world class super computers, today pitted its Watson computer system against the two of the most skilled contestants to ever appear on the TV game show Jeopardy. Watson is a technological leap in artificial intelligence. It’s designed to recognize the forms of human language contained within Jeopardy questions. Then, using only the data stored in its 15 trillion bytes of memory (it’s not connected to the Internet), it formulates answers to these questions, as well as deciding if and when to “buzz-in” to state the answer. Fifty IBM Power7 servers provide the computational muscle to perform these tasks in fractions of a second.
The introduction to Watson below is really impressive to watch. If this is truly a representation of the current state of AI, the dreams of the science fiction writers may soon become reality.