Let’s have a look at what’s happening to the chatbots’ world. Recently a huge transformation occurred: you may have heard of “GPT-3” (now GPT-4). What’s it? Why is it so revolutionary? Microsoft adopted it to make it easier to find results on Bing. Students all over the world are using it to do their homeworks…Here some info to shed some light on Chat GPT and the chatbots’ universe.

But first: what’s a chatbot?

If you know what we are talking about, then skip to the next paragraph. Otherwise, here’s an explanation for you. 

Basically, a chatbot is a software that interacts with the users and simulates a human dialogue. It works both with oral or written conversations to help users receive information, book tickets, set up an appointment… just as if they were talking with a real person! 

You surely have already chatted with a chatbot: for example, to order a pizza by phone or to ask for info on a website (e.g. writing your question on the chat opening on a specific page).

A chatbot’s speech can be more human-like or robot-like. What’s important is that it can manage any kind of request: this is the real challenge for any conversational designer. 

To build an effective conversation writers have to come up with any possible question users may ask and write the proper answers. There are a lot of platforms to do it, but creating a working conversation flow isn’t as easy as it sounds: there are always technical and logical difficulties, such as finding an equilibrium in the tangle of flows. 

However, the biggest limit is the human mind. It’s impossible for a single writer to imagine all possible combinations of words and questions a user can make. If flows don’t include some intents, the chatbot won’t be able to answer properly and, eventually, it’ll enter a loop of wrong outputs. 

The unexpected revolution

At the end of 2022 a voice spread: a new revolutionary LLM (Large Language Model) turned the tables of chatbotting, called GPT-3 (Generative Pre-trained Transformer, third version. Recently fourth version). 

GPT is a machine learning model created by OpenAI. It’s trained on any kind of Internet data and it’s able to generate texts starting from simple inputs. “Simple” means that no coding skills are needed, it’s enough to write sentences as in an everyday conversation. It can be used to create images, code, texts, music… and chatbots.

ChatGPT is the specific model to build them. I asked it the main difference with GPT-3 and here’s the answer:

“ChatGPT4 is a version of GPT-3 specifically designed for natural language processing (NLP) applications. It has been optimized for conversations and is able to generate personalized and interactive responses to user input. GPT-3, on the other hand, is a large language model developed by OpenAI. It is capable of generating human-like text, based on training with a large amount of data and can be used for various NLP tasks.”

Why are we talking about revolution? With ChatGPT, writers and developers don’t have to build complex conversation flows. Their job becomes so much easier! Not only for the coding issues, but also because ChatGPT can understand the context of the conversation and consequently give the proper answer. 

We are still at the beginning

ChatGPT still has to improve. There are some limits employing it in chatbots:

• it makes up answers when it isn’t trained on a given topic
• it isn’t alway easy to control what it says (it could give controverse answers) 
• to train it on specific topics (e.g. to create an help desk chatbot for an app) coding skills are required, since it wouldn’t be able to find the answers on its knowledge background 
•  it’s pre-trained, which means it isn’t constantly learning

Despite these limits, Microsoft has already integrated it on its browser Bing, betting on its huge potential.

What’s next? GPT-4 opens the door to a deeper connection between the AI space and the real world, redesigning the meaning of “Intelligence”.