What is the difference between computer and brains? Can computer memory be compared with the humanbrain? The first scientist whoexperimented to answer these questions related with the differences betweenmachinery and humans was Alan Turing, one british scientist. The first idea for measuring artificial intelligencewas famously becoming known as the”Turing test”.
In the 1950 paper, computer machinery and intelligence Turing propose the following game. A human judge had a text conversation withunseen players and evaluates their responses. In order to pass the test acomputer must be able to replace one of the players without changing results.In other words, a computer would be considered intelligent if the conversationcould not be easily distinguish from humans.
Turing predicted that by the year2000 machines with 100 MB of memory would be able to easily pass his test, buthe made a jumping on. Even though today computers have far more memory thanthat, he has succeeded and those thathave done well ,focused more on finding clever ways to full judges thenusing overwhelming computing powers. Thoughtit was never subjected to a real test, the first program that was thought to success was called”Eliza”. With only a very short and simple script, it ménage to lead astray many people, encouraging them to talk more andmirroring their own individual questions back at them. Another paw was named “Harry”.This script took the opposite transaction by imitating a paranoid schizophrenic who caped running the conversation back to his ownpre-programmed obsessions. Their success in fooling people pointing out one impotenceof the test. Humans regularly predicate intelligence to all series of things that are not actually intelligent.
Nerveless, annual competition has madethe test more nominal with judgesknowing that some of their conversation’s partners aremachines , but while the quality has proved that many chatbot programmers have used similar strategies to Eliza andHarry. 1997’s winner of competition “Catherine”, could carry amazingly focusedin an intelligence conversation, but mostly if the judge wanted to talk aboutBill Clinton and the more recent winner Jugine Goustman was giving thepersona the 13 years old Ukrainian boy, so judges interpreted its non isolationby using a clear grammar as language andcultural barriers. Meanwhile, others programs like “cleverbot” have taken an differentapproach by analyzing very big data bases and real conversation todetermine the best responses.
Some machinery also stored memory of last conversations in order toimprove over time. But, while cleverbot’s individual responses can sound likehuman, it is lack of consistent personality and inability to deal with brandnew topics which have been predicting that today’s computers would be able to discover planets and stars, perform difficult medicine operations and solve the most difficult math equationsbut, still struggle with the most basic small talk?! Human language turns outto be an amazing complex phenomenon that can not be captured even by thelargest dictionary. Chatbots can be befoul by simple exclamatory like:”Umm..” or complex questions with no correctanswer.
In a simple conversation, composite sentences like:” I took the juiceout of the fridge and gave it to them , but forgot to check the date”, requires a wealth of underlining knowledge and intuition. It turns out thatsimulating a human conversation takes more than just an increasing memory andprocesses. As we get closer to Turing’s goal, we may have to deal with allthese big questions about consciousness.