ChatGPT (Generated Pre-trained Transformer) is a language Chabot which was programmed by OpenAI (an open artificial intelligence research institute) and was launched on November 30th of 2022 and in just a duration of 5 days till the 4th of December 2022 it attracted over a million users where the famous streaming platform “Netflix” gathered 1 million users in 41 months, Twitter 24 months, Facebook 10 months, Instagram a month and ChatGPT only 5 days.
ChatGPT is a program which is built to answer a user in a human like manner it is trained on language text from the internet and uses a learning method called Transformers which basically enables the Chabot to generate answers in a human like manner; it can create texts that engages the user in a conversational/exchanging dialogues like setting.
ChatGPT is more efficient than other language models due to its Transformer technology which enables it produce longer texts.
It is use super convenient to use and to use it you just have to sign up by simply searching ChatGPT on the web page.
PURPOSE AND CAPABILITIES OF ChatGPT, WHAT CAN IT DO:
Generating text: ChatGPT can easily generate text of any sort for you; it can generate text for a new topic, summarize a piece of text for you, create a story, and complete a text and more.
Managing dialogues: ChatGPT has the capability to engage and mange a human like conversation by keeping up the tone of the conversation which makes it seems as real life human interaction.
Providing information: ChatGPT can provide a vast range of information on diverse topics and can answer to all of your questions in a chat like manner.
Tone analysis: ChatGPT has this capability where it analyses the emotions in a text which is great if you want to avoid certain type of feelings added to your text.
Translation: It can translate text from a language to another language based on the data it was trained on.
Purpose of ChatGPT: The data this Chabot has been trained on aims for suitable answers according to the user’s need in a conversational manner which makes fun and easy.
It can remember what it was asked before, what the user said in the previous conversations and it also trained to decline inappropriate requests.
This is what it replies with when its asked with such a question 
Development of ChatGPT:
ChatGPT has been trained and developed on a large handful of data gathered from various sources from books or the internet, the process of its training and development is mentioned below:
COLLECTION OF DATA: The step to start it all was the data collection in which the data was collected from various sources like websites, books internet and many more and fed to the model.
PRE-PROCESSING THE COLLECTED DATA: This step involved the removal of the duplicated data and also any sort of irrelevant information that was added to the collected data which basically cleaned the collected data.
MODEL DESIGN: This step involved using the transformer architecture and defining the model’s architecture.
TRAINING: The model is trained on the pre-processed data in which the data was installed in the model and brought to perfection.
FINE-TUNING: The model was fine-tuned to provide high detailed and specific answers.
EVALUATION: The model was tested again and again to improve its performance, once it gave a satisfactory result it was launched as a Chabot which can generate human-like text.
IMPACTS OF ChatGPT:
ChatGPT has took over the world by a storm, it can write poems, help programmers to write codes or it can compose songs or poems so it is and it will be creating an impact on the world.
It was created on GPT-3 which is the most powerful and the largest language model ever created.
It can help a lot of professions such as to help teachers in creating courses and so forth but it also puts a lot of other jobs at risk by taking over them like content creators or researchers.
This model sometimes gives false information which can create confusion in many ways and it can also become handy for cybercriminals.
ChatGPT is one of those programs which upon feeding it data will improve its performance even more which will put a greater impact in the near future as it’s still quite new.
LIMITATIONS OFFERED BY ChatGPT:
ChatGPT is changing everything but it still has its own limits just like everything else do, let’s take a look at some of them:
LACK OF EXPRESSION: ChatGPT is unable to have any sort of expressions just words which makes it unable to produce content which shares an intensity of any kind of expressions.
REAL LIFE INFORMATION: It also fails to give you live updates for example about the current weather.
LIMITED KNOWLEDGE AND UNDERSTANDING: The device is only loaded with the data till 2021 any time after 2021 will be more likely that the model will give you answers with less accuracy, and it just simply has less knowledge about specific domains like a language or a reference barrier.
MISUSE OF THE TECHNOLOGY: Because of its text as a human like manner it could be used to cause ill to someone or to spread hate.
IMPACTS ON JOBS: Further advancements in ChatGPT can cause job displacements which were previously done by humans which can cause an increase in the unemployment rate which is too risky.
HALLUCINATIONS: Because of ChatGPT being a language model it will inevitably provide answers which are hilariously wrong these are called hallucinations, it will sometimes just to do that and this problem is faced by almost every AI Chabot. OpenAI published a warning about incorrect output:
“ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.”
DETECTABLE AS NON-HUMAN: ChatGPT is highly organized and too wordy on top of that it uses a specific way of writing its text which makes it non-human and easily detectable,
LACK OF ACCURACY: The text ChatGPT provides is by identifying the word before some other word and not as a topic which makes it inaccurate.
OVERLY DETAILED: When you just want a direct answer to a question such as in a medical context it will still provide you with a very detailed and a comprehensive answer when a direct answer is preferred.
UNNATURAL: The model sometimes overlooks a topic and gives a feeling of being unnatural which also differentiates it to humans who are divergent unlike ChatGPT.
BIAS TOWARDS BEING NEUTRAL: Biasing towards positivity or neutralism is helpful but in cases with sensitive topics neutrality becomes an unwanted topic.
TOO FORMAL: Because it uses proper punctuation and all the right grammar the human like mannerism is neglected because humans tend to use a lot of slang in their daily talks and ChatGPT doesn’t get irony, sarcasm or human expressions.
CONCLUSION:
ChatGPT doesn’t have any personal opinions, biases or views and it’s trained to provide neutral and factual information based on the data it was trained on, it does not have any emotions and its goal is to provide helpful and informative responses to all of its users.
