If you are an active internet user then chances are that you already have had experience of talking to chat bots. Either by mistake or out of curiosity you came across one of these computer programs which were designed extra for maintaining a real conversation with humans. Such chat bots can be easily found on many entertaining sites, chat rooms, instant messengers. Hundreds of them are there to keep you company for a small talk or answer your questions on some really complicated topics.
However, the results of talking to chat bots are usually disappointing. Some of them are so poorly created that after a minute of a chat your “virtual” conversation partner will stop talking sense or will keep repeating one and the same phrase. There is only a really small number of programs which are able to keep the ball rolling. Their creators talk about artificial intelligence, and probably in the next few years it will be complicated for us to figure out who we talk to – a human or a bot.
History
The first, and probably, the most famous chat bot is ELIZA. The name refers to Bernard Shaw’s Eliza Doolittle from “Pigmalion” who was taught the language of the high society while the plot was unraveling. The program’s author, Joseph Weizenbaum, created in it 1966 after analyzing work peculiarities of psychotherapists. As we all know, some from experience, some from numerous TV shows, shrinks prefer to have the dialogue with patients in the mode of active listening which implies constant repeats and requests to continue talking. ELIZA communicated using the same model. In this case the chat focussed of asking questions instead of giving answers, the rest of the chat contained filler phrases.
The dull conversation got exciting every now and then when the chat bot was asking questions reacting to the key words in the speech of the chat partner. For example, if you said “My father hates me,” the bot would ask “Who else in the family does hate you?” Of course, there was no deep analysis of the words’ meaning. All the words, used by the chat partner were just a set of character data. However, ELIZA chat bot made a great impact on the future of chat, or chatter, bots, showing that it was possible to mess with the human language and to create some imitation.
1972 the psychiatrist Kenneth Colby wrote another program, his answer to ELIZA. His chat bot, PARRY, imitated communication style of a paranoid schizophrenic. A few times ELIZA and PARRY were given a chance to communicate with each other. Since that time making bots “talk” to each other has become one of the favorite time passing of chat bots’ developers.
Some time later, a number of computer games used the ELIZA’s algorithm, but it was way later, already in the 21st century, when perfect environment for chat bots was found. We are talking now about instant messaging. It all started with Kevin Fox’ AOLiza which again used the same algorithm as our dear friend, ELIZA. In a way it was a very successful experience since many IM users were accustomed to random chat partners and excessive use of bastardized language. Senseless answers of the bot did not discourage people and they tried to search for some meaning in them. Fox’ experiment was very important for the future development of similar programs. Nowadays you can find a lot of chat bots in each IM network, some of them were specially created to mention a certain brand in the conversation.
Testing for Humanity
It is supposed that the ideal chat bot should be able to pass Turing Test. There are even yearly competitions for such programs. The most well-known of them is Loebner Prize.
Turing test was mentioned for the first time by (as the name suggests it) Alan Turing in 1950 in his article “Computing Machinery and Intelligence.” The test is designed to testing the computer for having human intelligence. The idea was to see if computer can “think”. The process goes as following. A human judge is talking with two chat partners, one is human, and the other is computer. The task of the judge is to figure out who is who. If he/she can’t give a definite answer, then it means computer passes the test.
Chat Bots of the 21st century
Loebner Prize encourages developments of chat bots. However, till now there are not many chat bots that are close to leading a “human” conversation. There are two favorites as of now Jabberwacky, A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) and D.U.D.E.. They are constantly developing, learning new responses and trying to take advantage of new technologies.
Chat Bots in Business
A great number of companies that have active internet presence use the technology in their best interests. Chat bots are used as virtual assistants that can answer customer’s questions, help the companies to accomplish their marketing purpose, maintain their twitter accounts.
Future of Chat Bots
More and more developers try to get close to grasping the idea of artificial intelligence. For them there is no question whether they succeed or not. The question is when they will be able to create a computer program that will act, think and talk like a human. And what will be the consequences. Till then the developers will make more attempts in improving existing chat bots and creating new ones. How much time it will take them nobody knows. But better watch out, maybe your new best online friend is a bot.
But then again, maybe not.
No comments:
Post a Comment