Hello, in this sixth article I want to introduce you to CHATBOTS, chat bots or conversational bots.
In short, they are applications that emerged in the 1960s, and that simulate having a conversation with a person by giving automatic responses, which are previously established according to the inputs made by the user. These bots, also known as expert systems, use case-based reasoning or CBR: Case Based Reasoning.
Usually, the conversation is established by text, although there are also models that have an interface that allows voice interaction through text-to-speech (TTS) and voice-to-text (TTS) converters, providing more realism to the interaction with the user and helping to reduce response time. This is where CHATBOTS have made notable progress in recent years incorporating Artificial Intelligence techniques such as Natural Language Processing (NLP) and Machine Learning (ML).
To establish a conversation, easily understandable and coherent phrases must be used, although most chatbots do not fully understand. Instead, they take into account the words or phrases of the interlocutor, which will allow them to use a series of responses prepared in advance. These are capable of recognizing the way in which a sentence is formulated thanks to a series of pre-established comparative patterns, and in this way, based on the different variables of said sentence, they present a corresponding response. In this way, the bot is able to follow a conversation with more or less logic, but without really knowing what it is talking about.
The main objective of CHATBOTS is to improve customer service (in the field of marketing), that is, to generate quick and concise answers to common questions from users. In addition, to send information / news and advertising automatically from the company. In the educational field, its main objective is to maintain fluid communication between students, family and school.
In relation to quality control, there is the Turing test to determine the level of understanding of the bot. The Turing test is the industry standard that identifies whether the bot has the ability to generate intelligent behavior with the user. Therefore, the development of the test allows the chatbot industry to be of quality and to be framed as intelligent assistants.
In a previous article we talked about GPT-3, an Artificial Intelligence developed by the company Open AI, based on an autoregressive language model that uses deep learning to produce texts that simulate human writing. GPT-3 always has an answer, but is never “aware” of whether or not its answer makes sense. It predicts the probability that something is correct, but cannot reason about whether or not it is.
The way GPT-3 works is its main strength and weakness. When a text is injected into it, the model predicts what should come next thanks to all the parameters it has ingested, then, from the text that the same model predicted, it generates the following words, and so on until it reaches its limit. If at any time the text it generates loses consistency and meaning, GPT-3 will not know because it does not understand the context, it only generates the following according to a probabilistic model.
GPT-3 is being used in the new generation of CHATBOTS, and is presented as an alternative with great potential for the future. Being a particularly good system for generating long texts that maintain a certain coherence between sentences, we will see more and more models of this type used to generate texts semi-automatically. And not only for conversational assistants of all kinds, but also to help in the writing of news or other types of reports, and even in the search for information. The only drawback is that since it lacks abstract reasoning, it is easy for it to make mistakes that a person would never make. In other words, although it passes the Turing test with good marks, when you have been talking to this AI for a while, you end up noticing that it is easy for it to generate absurd answers if the questions start to be crazy, although grammatically correct.
With the arrival of the Metaverse, it is normal to want to integrate CHATBOTS and other Intelligent Assistants into Automatic Avatars that can help us in the New Immersive Experiences that will make up the next Evolutionary Phase of the Internet.
When a CHATBOT not only has the capacity for intelligent conversation, either by text or voice, but also acquires avataric corporeality, we call it a Digital Human (Digital Human).
But this is only the initial stage of a Digital Human, since the concept goes much further. The main idea is that Artificial General Intelligence (AGI) solves its main weakness, which is at the same time what differentiates us humans from machines: the lack of creativity. It is as if we put a human brain inside a machine, so that artificial intelligence is capable of doing everything that we humans feel and do.
It won’t be long before that moment arrives. Digital humans will completely transform the relationship between companies and customers, by being able to offer ultra-personalized experiences in a totally humanized way. The most important influencers will be Digital Humans, whose capacity for work and virilization will be unmatched, and even celebrities will have their Digital Twins, who will work with the same technology to interact with their fans as if they were themselves, since these doubles will behave exactly the same as they. With the arrival of Digital Humans, the world will be safer if we educate them with humanist ethics, goals and empathy, which opens up great development potential for us, for example, training digital doctors for areas of the planet where they now do not have access to the health.