DACS General Meeting
June 2012

Meeting Review:
Bruce Preston—Artificial Intelligence

By Richard Corzo

At our June 5th general meeting it was time to learn a little about computer science. Inspired by last year’s IBM Watson Jeopardy! challenge, longtime DACS member, Access SIG leader, and past presenter Bruce Preston, laid out the path that led us to the current state of artificial intelligence. How can a machine mimic the responses of a human?

He started out by explaining electrical circuits and how they can represent the 0s and 1s in a digital computer. Special circuits representing ANDs and ORs lay the foundation for representing logic in a computer, and for building some basic arithmetic operations.

Bruce gave us a little computer history, explaining the term “bug” originated with a World War II-era computer that was encountering errors. They literally found a moth in a relay of the computer that was causing the problem, and thus had to “debug” the computer to get it working again.

One challenge was reducing the size and increasing the speed of computers, as the first ones filled entire large rooms. Vacuum tubes  were replaced by transistors and then integrated circuits. Eventually an entire processor fit on one chip.
The next hurdle was software. The earliest programs were hard-wired into the computer, but that was replaced by programs stored in the computer’s memory. The simplest programs are a series of instructions, but more complex programs are built using reusable components. Algorithms are designed to perform a particular complex task, such as searching for a particular string in some text.

British mathematician Alan Turing devised a test for artificial intelligence. Without seeing them, if a judge cannot distinguish a human’s answers from a computer’s, then the computer is deemed to have artificial intelligence.
Two more developments were need toward this goal. The first was natural language processing, of which the Eliza program was an example. She gave answers and posed questions that mimicked that of a psychoanalyst. The second development was rules-based systems. IBM’s Deep Blue was a rules-based system that was eventually able to beat a human chess champion.

Combining natural language processing with a rules-based system, IBM developed the Watson computing system which was able to compete with Jeopardy! contestants last year.

As a final example Bruce along with iPhone owner Richard Corzo demonstrated the Siri assistant on the iPhone. It combines speech recognition (with no prior training to recognize the iPhone user’s voice) with a rules-based engine located on some Internet-based servers. It had no trouble answering who won the Super Bowl in some particular year. The audience made a few suggestions as to what to ask Siri, and unfortunately Siri heard them all at once, but picked one of them to provide an answer.

It’s amazing to think how far we have come. And we should give Bruce a special thank you for moving his presentation up a month to replace our original June speaker who couldn’t make it.

 


Click Here

DacsGear!
Mugs and more, visit CafePress to order

 
 
© Danbury Area Computer Society, Inc. All Rights Reserved.
Web Site Terms & Conditions of Use