Chatbots, also known as avatars, chatterbots or lingubots, are computer programmes that use artificial intelligence methods and techniques to carry on a conversation with a user in natural language. These systems are usually placed on the websites.
A chatbot can talk about almost any subject and, if equipped with a company- or institution-specific knowledge base, can be transformed into an authentic Virtual Advisor in the role of salesman, promoter, service technician or customer service employee. Their task is to provide customers with information on the products and services offered, the specifics of the company’s operations, or e.g. business conditions. They often also have entertainment (e.g. in games) or advisory functions (e.g. a virtual credit advisor or doctor). The conversation with the chatbot itself is usually in the form of a chat via a website.
The prototype of all chatbots was ELIZA, created by Joseph Weizenbaum back in the 1960s. ELIZA pretended to be a psychotherapist and allowed a simple dialogue with the user. The mechanism of ELIZA was extremely simple and had little to do with artificial intelligence. In the 1980s-90s, chatbot technology developed rapidly. Today’s chatbots have gigantic knowledge bases, can associate certain information and learn based on the user’s input.
Looking at it from a a technical point of view, a chatbot is nothing more than a computer programme based on a certain algorithm and simulating a conversation. It is an application that presents the possibilities of conversational Artificial Intelligence. Such technology enables communication between a user and a website, online shop or any computer application in natural language. The user conducts a dialogue with the chatbot by asking questions in natural language and the system responds with context and knowledge base.
It is a kind of virtual assistant, capable of performing a wide variety of functions. The chatbot’s task and the main difficulty it has to overcome is to recognise language structures so that it can be communicated with like an intelligent human being. Often chatbots are also endowed with a specific personality, a character, so that they simulate human behaviour even better.
This technology can be used to, among other things:
- make a website more attractive and reduce helpline traffic,
- creating interactive services, e.g. technical support,
- create information systems for employees or partners of a company,
- create advisory services in specific areas,
- making computer games more attractive, etc.
Chatbots currently being developed use mixed methods – machine learning, hidden Markov models, semantic networks are used alongside natural language processing.
The better the simulation of the conversation, the more intelligence is attributed to the system. Efforts here are directed towards, through an application they have developed, a full Turing test (A way of determining a machine’s ability to use natural language and indirectly intended to prove its mastery of human-like thinking skills. The test looks as follows: A judge – a human – conducts a conversation in natural language with the other parties. If the judge is unable to reliably determine whether either party is a machine or a human, then the machine is said to have passed the test. It is assumed that both the human and the machine are trying to pass the test by behaving as close to human as possible; see https://glivia.com/turing-test-imitation-game/ ). Many researchers still believe that it is language that is the barrier that, if overcome, will grant a machine the actual ability to think. Although passing the Turing test is still an illusion, there are claims about its inadequacy to determine the intelligence of a computer.
See what a conversation with our chatbot looks like: https://glivia.com/chatbot/