Instead of having direct communication with a live human agent, a chatbot is a software programme used to perform an online chat conversation using text or text-to-speech. Designed to convincingly mimic how a human being will behave as a conversational partner, chatbot systems usually need constant tuning and checking, and those in development are unable to converse or pass appropriately.
For different uses, including customer support, request filtering, or for information processing, chatbots are used in dialogue systems. While some chatbot implementations use elaborate word classification processes, natural language processors, and advanced AI, others merely use standard phrases retrieved from a related library or database to search for general keywords and produce responses.
Via website popups or by virtual assistants, most chatbots are accessible online. They can be grouped into use categories that include: commerce (e-commerce via chat), education, culture, finance, fitness, news, and productivity.
In 1950, the popular paper “Computing Machinery and Intelligence” was written by Alan Turing, which introduced what is now called the Turing test as an intelligence criterion. This requirement relies on the capacity of a computer programme to impersonate a person in a written dialogue with a human judge in real time to the degree that the judge is unable to consistently discriminate between the programme and a real human on the basis of the conversational material alone.
The notoriety of the proposed test by Turing generated considerable interest in the programme ELIZA, published in 1966 by Joseph Weizenbaum, which appeared to be able to trick consumers into thinking that they were talking to a real person. However, Weizenbaum himself did not say that ELIZA was actually intellectual and addressed it rather as a debunking exercise in the introduction to his paper:
Artificial Intelligence [In]… Machines are designed to act in marvellous ways, often enough to dazzle even the most professional observer. But if a single programme is unmasked, once it describes its inner workings… Its magic is crumbling away; it is exposed as a mere series of processes… The analyst says to himself “I could have written that”. He moves the software in question with that thought from the shelf labelled “intelligent” to that reserved for interest.
In addition to identifying key words or phrases within the feedback and outputing corresponding planned or pre-programmed answers, ELITZA’s central organisational strategy (copied from chat designers) is to make the dialogue proceed in a seemingly relevant manner (for example by addressing every entry containing the word ‘Mom’ with ‘TELL ME MORE ABOUT Your FAMILY ‘). ELIZA demonstrated that this is remarkably easy to establish, as man’s judges are prepared to benefit the doubt when conversational reactions can be understood as “intelligent”
Interface designers have come to appreciate that people can be used for useful purposes to view computer output as truly chatty — even though it is simply built on very simplistic pattern matches. Many users tend to communicate with human-like applications and that allows chatbot-style techniques, a potentially useful function in interactive networks, so long as this knowledge is reasonably simple and is categorised as predictable. For example online support systems may use chatbot strategies to define the help area required by users to offer a “friendlier” interface rather than a more traditional search or menu structure. This way of usage raises the possibility of converting chatbot technology to that marked “shelf … reserved for curios” from the “genuinely useful computational methods”
There is so much more to learn regarding chatbot.
Find more articles at finsliqblog.com