Search artificial intelligence, capable of a friendly conversation, was not a little long. When Alan Turing, the father of modern computing systems, presented his famous test, which was to show the true intelligence in a computer program, he set a very ambitious goal for iron. If the computer could convince a group of judges-the people they’re talking to a man – if he could keep the conversation going – this would indicate the presence of artificial intelligence has evolved to such an extent that was indistinguishable from human.
This glove was abandoned in 1950. Still no computer program could not pass the Turing test. Private leadwire doesn’t count.
Why jepodise? Because they probably think failures than achievements. For example, Joseph Weizenbaum in 1966, when computers were programmed with punch cards great, developed the software for processing of natural speech – ELIZA. “Eliza” was a machine that was supposed to keep the conversation going, pretending to be a therapist; it is possible to talk today.
To talk to “Eliza” is a bit strange. She often paraphrases what you said, so, for example, if you say “I feel depressed”, she might reply, “You came to me because you feel depressed?”. When she did not understand what you say, she answers “yeah” or “tell me more”.
In the first lines of dialogue, especially if you treat her as your doctor, “Eliza” can be very convincing. When Weizenbaum noticed it, it is a little unsettling: people were ready to see the algorithm more than it was humane. Soon, though, some subjects realized that they were dealing with the car, they opened her your deep feelings and secrets. They poured out their soul to the machine. When the Secretary of Weizenbaum talk to “Eliza”, even though she knew she was talking to the program, she still insisted, Weizenbaum left the room.
Part of an unexpected reaction, which was called ELIZA, can be caused by the fact that people are more inclined to open the car, knowing that no one will judge, even if the machine won’t be able to say anything or do anything to help. The Eliza effect was named that in honor of this computer program: people tend to endow machines with human features, or to think of them as people.
He Weizenbaum, which subsequently became deeply suspicious of the impact of computers and artificial intelligence on human life, was amazed that people were willing to believe that his script was a man. “I could not have imagined that a very short acquaintance with a simple computer program will lead to such delusional thinking in quite normal people.”
Effect Eliza could alert Weizenbaum, but he’s intrigued and fascinated others for decades. You may have noticed it on myself, talking to the AI like Siri, Google Assistant or Alexa — these short answers seem too real. In your right mind you know what you’re saying with a large piece of code stored somewhere in the ether. But subconsciously you feel like you’re talking to a person.
Further — more. Facebook opened up its Messenger and gave the software people and companies, so that they could create their own chatbot. Processing natural speech progressed by leap and crawling from the 1960-ies. Today, you can find friendly chatbots like Mitsuke, who recently won the loebner prize, which is awarded to machines that are ready to solve the Turing test. She will respond to everything you wrote. The idea is simple: if you already have an app for ordering pizza, why not order a pizza, asking, for example, a chatbot like an old friend?
Startups like Semantic Machines hope that their virtual assistants will be able to interact with you, like secretaries, with the ability to get information from the Internet. Soon they will be everywhere.
But the people who make chatbots — both social and commercial vein, face a common problem: people may subconsciously accept chatbots for people and get upset when they can’t hold a normal conversation. Frustrated with the misunderstanding often associated with inflated initial expectations.
Still no machine could learn to understand context — to consider what was said earlier, to refer to it and respond based on the current situation of the dialogue. Even Mitsuke will often try to remember the topic of conversation after a few lines of dialogue.
It is clear. The conversation can be multi-faceted and complex. Everything you say could be hundreds of answers that will make sense. When you create additional layers of conversation, these factors multiply, multiplying, and all kinds of variations of dialogue. It’s harder than chess or go.
But this does not prevent people from attempting to create a new chatbot. Recently Amazon launched a competition for the Alexa Prize in which the winner of AI will be awarded a prize of 500,000 dollars and another million dollars on top, if the team will be able to create a “social bot” that can hold a conversation with a human user within 20 minutes on a variety of topics.
Among the designated topics — science and technology, politics, sports and celebrity gossip. Was recently announced finalists: chatbot University in Prague, Edinburgh and Seattle. The finalists were selected in accordance with user ratings Alexa.
Narrow down your conversation to a specific range of topics chatbot starts to skillfully avoid the problem of context. It is much easier to simulate a conversation, which relates to the limited scope of topics.
The development of machines that can support almost any conversation with someone in person, can be challenging. Perhaps she will require artificial intelligence to a complete solution than previously used approaches with recorded responses or neural networks that associate inputs with answers.
But the car, which will conduct a substantive dialogue and where people can enjoy, is sure to be. Alexa Prize winner will be announced in November. The Eliza effect means that we will rely on machines earlier than thought.
Turing dreamed of machines for communication. When his dream will come true?
Ilya Hel