The mysterious history of ChatGPT

chat bot, chatGPT

Despite the great media boom that we are currently experiencing with tools such as ChatGPT, chatbots and conversational AI come from afar and have had to overcome big bumps to get to the brilliant news we see now.

Traditional chatbots and those that make use of conversational AI have been around for a long time, but only in recent years have they gained real popularity among users and companies with great tools like ChatGPT breaking all the schemes.

For the most part, this shift in consciousness towards conversational AI and more large-scale chatbots has occurred with the advances in artificial intelligence and machine learning that we are continually seeing.

ChatGPT has been in development for several years and has undergone numerous updates and improvements, making it one of the most advanced chatbots available today.

However, it seems that the birth of this type of technology starts from here, when its fame has grown like foam, but conversational AI and chatbots come a long way and have had to evolve a lot to get to where we are today.

  • Conversational AI vs. traditional chatbots
  • The dawn of conversational AI
  • Is ChatGPT really different? 
  • The future of conversational AI and chatbots

Conversational AI vs. traditional chatbots

Before delving fully into the subject, we must clarify that traditional chatbots are quite different from the current ones that make use of what we have already mentioned as conversational AI. This has to do with the tools and programming that allow a system to imitate and carry out conversational experiences with people.

A chatbot is a program that can (but not always) use conversational AI. It is the program that communicates with people. Conversational AI powers chatbots. But not all of these use conversational AI.

AI-powered chatbots use conversational AI to understand and converse with you. Machine learning will allow chatbots to remember the things you have said to them and allow a more realistic conversation to flow.

Natural language processing allows chatbots to understand a broader range of input and determine the intent behind your messages, and intelligent analytics allows chatbots to make recommendations based on our past records and interactions.

The dawn of conversational AI

It was in the 1950s that Alan Turing hypothesized that a computer program could interact with humans. 

The first chatbot software ever developed, ELIZA (1966), is a computer program that uses natural language processing to mimic the speech of a psychotherapist. This is the first time that a human-like conversation was made possible with a chatbot.

ELIZA paved the way for many other conversational experiments, but she was unable to generate context from the conversation and had limited general knowledge.

After this came some others with better capabilities, although mobile phones have been the catalyst that accelerated the development of artificial intelligence . In the early 2000s, they sought to adapt websites to other smaller screens, they struggled to improve graphic designs and functionality, as well as wondering if there could be a better interface to improve the user experience.

With the development of different mobile applications, the first chatbot technology that emerged in 2001 allowed users to retrieve information about prices, sports results, movie times, yellow page listings, weather and news. 

The development of SmarterChild (2001) marks a turning point in the evolution of chatbot technologies. This took a step forward by connecting users with external information sources and also really got hooked. At its peak, SmarterChild reportedly had 30 million users.

Afterwards, many apps were developed between 2010 and 2016, with the most popular being Siri, Google Now, Cortana, Alexa, and Google Home.

In the early stages of the development of chatbots, basic NLP (Neuro-Linguistic Programming) methods were used to design them, as machine learning wasn’t exactly feasible at the time. 

Conversational AI got a lot better with the arrival of Transformer in 2017. These revolutionized the field of language by processing language in context. One of the bases, as we have said before, for a chatbot to go further in its conversations with humans.

His novel approach was also more computationally efficient, making it easier to scale up model parameters and training inputs. Not long after, Transformer-based models such as Google’s BERT and OpenAI’s GPT series of models were released with incredible performance.

Using the Transformer architecture, Google led several innovative experiments in conversational AI, developing models that excelled in conversations on many topics . LaMDA (from Google) in 2021 managed to improve all previous processes and was so good at simulating a natural language conversation that one engineer claimed it was clever.

And then of course came ChatGPT. This was tuned using human feedback on a Transformer-based GPT 3.5 series model.

OpenAI refers to ChatGPT as a “sister model” of InstructGPT and uses similar methods, but when training with human dialog datasets, as well as InstructGPT’s dataset translated into a dialog format, the result is much more conversational.

Is ChatGPT really different? 

It is certainly very effective. It provides very coherent and sometimes even intelligent answers to complex questions. Its ability to generate natural language responses is impressive, and it is able to maintain context (at least at a surface level) to subsequent questions. 

Now, what does this mean for the research and development of AI-powered chatbots? We’re done? ChatGPT is one of the most exciting developments in artificial intelligence in recent years and opens up a world of possibilities, from healthcare advice to customer service to virtual assistants. The potential applications are truly endless.

But while the possibilities are exciting, there are also some risks associated with ChatGPT. For one, it’s not always easy to distinguish between a real person and an AI-generated response . This could cause confusion for users or even be exploited if used for deceptive purposes. Also, if not used responsibly, it could lead to ethical dilemmas and privacy violations.

In general, the potential of ChatGPT generates enthusiasm for many and the possibilities it offers. If used wisely and responsibly, we can expect amazing advances in AI that will benefit us all.

The big point here to take into account is not the capabilities of ChatGPT, but rather that the conversational interface has returned to all media, positively, and not only through OpenAI. New apps are springing up in search, creative tools, customer service, and even social media.

The future of conversational AI and chatbots

Chatbots are an amazing field of research to undertake with literally endless possibilities to consider. Companies invest millions of dollars in research and it is really revolutionizing the world. 

The adoption of conversational AI improves personalization, as bots are able to have human-like conversations and remember user preferences, previous dialogues, context, and meaning. These are used to provide a rich and satisfying user experience and increase user engagement with the systems.

A big step in the research will be for companies to dive in to make chatbots focus on generating personas out of chatbots. 

A very advanced futuristic chatbot would be able to help humans in almost every field, from medicine, to customer service, to even education, and humans will mostly interact by talking. Of course, there’s still a long way to go, but there’s no denying that ChatGPT has resurrected old-fashioned chatbots with the potential of conversational AI.

Leave a Reply