Oyez Oyez, we meet this week for an article about the history of the chatbot, from birth to today. Before you start a few keywords to arouse your interest: Test Turing, Eliza, MIT, Siri and Alexa. Keep reading to understand the link between them.
Plan de l'article
- Definition of Chatbot
- History of the chatbot
- — It started with “Hello. Why are you coming to see me? Then, depending on the user’s answer, she tried to find a question in the extension. — Another case: the user asks her a question she answers her questioning the origin of this question: “Why this question? — And if the interlocutor pronounced a sentence containing the word computer (see search for keywords as stated in the definition), she asked, “Do you say that because I am a machine? The following chatbots are based on Eliza such as PARRY created in 1972 who interacted with her then ALICE in 1995 etc… The chatbots have since become more and more numerous, among the best known are Siri and Alexa. What is the best chatbot?
- Fun Fact
- The word of the end
Definition of Chatbot
A chatbot is a program that tries to have a conversation with a person for a few minutes whose purpose is to give him the impression of addressing a human.
Read also : Why use a chatbot?
We might then think that the program is supposed to understand what the person tells him, but this impression is false, most conversational agents are not designed to understand.
They spot keywords or expressions that are then called triggers to find the answer in the database. It works to a certain limit, the conversation is more or less intelligent, and does not require understanding what they are talking about.
Read also : How to become a good web editor?
To improve the flaws of this technique, namely to associate “how does it work? “How does the software work? a system based on both the keyword recognition method described above and a system of word recognition and language analysis must be used.
By launching the second system on the sentence stated by the interlocutor, the language analysis will try to find the information necessary to answer the question, if there is no match then the chatbot will fall back on the keyword method.
That’s why we need to differentiate two types of chatbots:
- — simple bots, based on the keyword method and
- — smart bots, which use language analysis based on natural language comprehension technology.
History of the chatbot
Before talking about the first best-known chatbot, namely “Eliza”, let me introduce her ancestor to you.
The word ancestor might seem strong to you but if I tell you that Eliza was created in 1964 and the chatbot I’m talking about dates back to 1780, you understand me better.
The man behind the first chatbot research is none other than a French abbot, Abbot Mical, he developed two copper speaking heads, who were able to pronounce four sentences by awkward reproducing the human voice, giving the impression that they had a conversation.
Subsequently there was Eliza as mentioned above that was created by a professor at MIT, Joseph Weizenbaum. Its operation was quite simple:
— It started with “Hello. Why are you coming to see me? Then, depending on the user’s answer, she tried to find a question in the extension. — Another case: the user asks her a question she answers her questioning the origin of this question: “Why this question? — And if the interlocutor pronounced a sentence containing the word computer (see search for keywords as stated in the definition), she asked, “Do you say that because I am a machine? The following chatbots are based on Eliza such as PARRY created in 1972 who interacted with her then ALICE in 1995 etc… The chatbots have since become more and more numerous, among the best known are Siri and Alexa. What is the best chatbot?
But then how could one evaluate the capabilities of a chatbot?
This is where the Turing test comes in: it is intended to test the category of chatbots that want imitate humans.
Set up by Alan Turing in 1950, this test works as follows: a jury talks with 2 users via interposed screens, one is a chatbot, the other a human. At the end of the test, the jury designates the one he thinks is a robot.
The flaws in the test are as follows: if the chatbot’s goal is to answer a customer’s questions but cannot simulate a certain humanity, then it will not validate the test while it can answer questions for customers very well.
After this test began, the Loebner Prize was established in 1990 by Hugh Loebner to award the most human computer program. Some chatbots won this prize: for example, the ALICE chatbot quoted earlier won the prize in the following years: 2000, 2001, and 2004. The latest winner is the Mitsuku robot.
One should ask whether the Tuning test is still relevant because according to research conducted by MIT, the more the chatbot seeks to look human, the more the jury is wary of it.
A little fun fact before you leave?
After Amazon released the Alexa Assistant, the number of girls called after that first name dropped by half from 2015 to 2018.
The word of the end
Robots, although their evolution has been quite remarkable over the last 20 years, have nevertheless not finished improving and surprising us.
Today, experts agree that chatbots have a place in some of a human’s daily tasks but cannot yet replace it, a constant we share at TeamBrain.
Do not hesitate to follow us on social networks to not miss our next articles that will hold a weekly cadence!
Linkedin facebook twitter Youtube