
Since the release of the bot, ChatGPT in November, users all around the world have been casually experimenting with it by asking it to generate responses. It has taken the world by surprise and experts by concern, for its accurate and fast way of generating various kinds of responses. Even Google has now announced the launch of its own AI-based chatbot, Bard.
But before these chatbots start becoming a common part of our life like other AI tools, let’s clear our basics first.
ChatGPT as you all know is a chatbot. A chatbot is basically a bot (robot) that chats, it can either be interactive or programmed. It is built by OpenAI and is made on top of its GPT-3 family of large language models. Silicon Valley players Elon Musk, Sam Altman, Peter Thiel, and LinkedIn co-founder Reid Hoffman founded it. However, following the creation of ChatGPT in 2015, Elon Musk resigned from it to avoid conflicting developments with his other company, Tesla.
It was launched as a prototype on 30 November 2022 and garnered a lot of attention for its detailed responses and articulate answers across many domains of knowledge.
GPT in ChatGPT stands for Generative Pre-training Transformer. As the name suggests, Generative is a model that can generate text. It is a family of language models that are generally trained on a large corpus of text to generate human-like text. The word Transformer is used here to suggest that it has built several blocks of its module on the basis of transformer architecture.
All these collectively help in generating various natural languages processing tasks such as text generation, language translation and text classification. Pre-training denotes the initial training process on a large text material where the model learns to predict the next word in a passage.
It might perform better than many of the other AI-based tools and can sometimes even surpass humans in generating responses like essays, scripts, coding etc. But it comes with its own limitations.
1. The responses that it generates are very formal and lack expressive language. It lacks sarcasm, analogies, and comedy because obviously, it is not human.
2. It just summarises and lacks depth on the subject or matter. On the other hand, humans create thorough information and also draw from subjective views and personal experiences.
3. It is biased and sometimes even provides wrong answers. So it can be dangerous to trust this AI tool for your assignments and might get you in trouble.
4. It is literally a machine language that is just trying to imitate human language. It sounds very unnatural and is still undergoing training and development, unlike humans, who have been developing since time immemorial.