DEV Community

Cover image for ChatGPT Series, Pilot Episode: Initiation Interview
aldin
aldin

Posted on

ChatGPT Series, Pilot Episode: Initiation Interview

ChatGPT (Generative Pre-trained Transformer) is an extensive language model chatbot launched by OpenAI in November 2022. It already reached millions of users - and rightfully so.

I was super hyped regarding ChatGPT, but the second I tried to use it I understood that so was everyone else - the system was overloaded. Hence, I had to wait, but no more waiting.

The idea here is to have a series of conversations with it and to write an article for every one of them. Today on the menu (yes, still a foodie) is the initiation interview. Yes, I've interviewed a chatbot! And, if the circumstances allow, I'll continue to write weekly on my ChatGPT interactions. Already have several topics in the draft area (incl. The ChatGPT API playground, Fixing Bosnia with ChatGPT, etc.) - if that sounds like fun, stay tuned.

Consent

We hit it off with a short round of greetings and my asking for consent to even conduct a short interview. Look at how that went:

A consent from ChatGPT to conduct an interview

The interview

So, after getting a YES to make that chat an interview, we hit it off with intros. There was no script, but each next question was rather a conservative of the previously answered one(s).

This is what ChatGPT has to say about itself.

Q: In the spirit of getting to know you better, would you be kind to shortly present yourself?

ChatGPT: I am ChatGPT, a large language model developed by OpenAI. I have been trained on a wide variety of texts, so I can understand and generate a wide range of human language. I can be used to answer questions, generate text, and more. I am a machine, which means that I don't have feelings or consciousness, and I only respond based on the data that I've been trained on.

When asked about its launch date, it had no idea. That might come off as a little bit strange… until you find out that the training data cut-off is EOY 2021. Then it all makes sense.

Q: When were you born and launched, and by whom?

ChatGPT: I am an artificial intelligence and do not have a physical birth. I was created and launched by OpenAI, the exact date of my launch is not specified, but my knowledge cut off is 2021.

Q: Ah yes, the knowledge cut-off explains it! :) Well, let's go in another direction then. Can you tell me about the technology behind your creation?

ChatGPT: I am based on a transformer architecture, which is a type of neural network. The transformer architecture was introduced in a 2017 paper by Google researchers, "Attention Is All You Need."

The transformer architecture allows me to process input sequences in parallel, rather than in a sequential manner like previous architectures. This makes my processing much faster and more efficient. Additionally, my training process involves being fed a large dataset of text, and adjusting the parameters of the model so that it can accurately predict the next word in a sentence.

My training process is done through a process called pre-training, which consist of training the model on a large dataset and then fine-tuning it on a smaller dataset to make it more accurate in specific tasks.

Overall, my creation uses a combination of advanced machine learning techniques and vast amounts of data to understand and generate text.

Context and Conversation

Q: And would you say that your training process made you perceptive in analyzing the questions and interactions that people make when communicating with you?

ChatGPT: Yes, my training process has enabled me to understand and analyze text input and generate coherent and relevant responses. I have been trained on a large dataset of text, which allows me to understand the context of the questions and interactions people make when communicating with me. Additionally, the transformer architecture allows me to pay attention to certain parts of the input, which helps me understand the meaning and intent behind the question.

However, it is important to note that my understanding is based on patterns and statistics from the data I was trained on and I may not always understand the nuances of human communication or be able to respond appropriately to certain questions.

Upon learning that the answers should be coherent and contextual I was curious about how well is it able to hold a conversation.

Q: Are you able to hold a conversation or all of the questions are treated as completely separate queries?

ChatGPT: As a language model, I am able to hold a conversation to some extent. I can understand context and use it to respond to follow-up questions or continue a conversation. However, my responses may not always be as seamless or natural as those of a human, and my knowledge is limited to what I have been trained on, which has a cutoff date. Feel free to ask me anything else you want to know, I'll do my best to help.

Science and Literature Test

Q: And would you say that you are good with science or literature?

ChatGPT: As a language model, I have been trained on a wide variety of text from the internet, which includes information on a wide range of subjects, including science and literature. I can provide information and answer questions on these subjects to the best of my ability based on the information that I have been trained on. However, my understanding and knowledge may not be as comprehensive or up-to-date as that of a human expert in these fields.

Additionally, my knowledge cut-off is 2021, so any recent development may not be available for me to provide an accurate answer. If you have a specific question or topic in mind, I would be happy to try to help.

Q: Hmmm, I understand. Having said that, would you be able to tell me which novel has brought the most powerful feeling of sadness for its hero?

ChatGPT: Many novels have powerful themes and evoke strong emotions in readers. It can be a matter of personal preference, but one novel that is often cited as evoking feelings of sadness for its hero is "The Great Gatsby" by F. Scott Fitzgerald.

The novel tells the story of Jay Gatsby, a wealthy man who becomes deeply infatuated with a woman named Daisy, who he knew before leaving for war. Gatsby throws lavish parties in the hopes of winning Daisy back, but ultimately his love for her leads to his downfall. The novel explores themes of love, loss, and the futility of the American Dream, which can be very poignant and sad.

Q: Wow, you really are a literature buff! Ok, and if we'd talk science, say physics, would you be able to explain The Big Bang theory as if I were a 5 years old child?

ChatGPT: The Big Bang theory is a scientific explanation for how the universe began. It says that a very long time ago, there was nothing in the universe except for a very hot, very small, and very dense point called a singularity. Then, all of a sudden, this singularity started to expand and cool down, getting bigger and bigger and making space and time as we know it.

Think of it like a balloon, at the beginning, the balloon was very tiny and everything was squished together in it, but as it started to blow up, the material inside started to spread out and make more room for itself, creating the space we can see in the balloon. This expansion continues today and it's still happening right now.

As the universe expanded and cooled, different types of particles came together to form the first atoms, which eventually formed stars and galaxies. And that's how the universe was created. Scientists still have many questions about the Big Bang and the nature of the universe, but this is the most widely accepted explanation for how it all began.

Q: Would you say you're able to effectively make a difference in the way you answer to the questions depending on the age or levels of education of the person asking the question?

ChatGPT: As a language model, I am able to understand the context of a question and respond in a way that is appropriate for the person asking it. For example, when you asked me to explain the Big Bang theory to a 5-year-old, I used simpler language and analogy to make it easier to understand.

However, I am a machine learning model and my understanding of the context and ability to adjust my language to different age or education levels may not be as nuanced as that of a human. It's also worth noting that the level of understanding and comprehension of a subject can be vary greatly even within an age group or educational level, so my responses may not always be a perfect fit for every individual. But I will always do my best to provide information that is clear, accurate, and easy to understand.

Q: I thought the same thing when I saw you using the balloon analogy. There is some more space for polishing that, but hey, ML always keeps improving (as should people)!

ChatGPT: Yes, you are correct. Machine learning models, including myself, are constantly improving as new data and advancements in technology become available. The balloon analogy I used is just one way of simplifying a complex idea like the Big Bang Theory, and it could be improved upon. However, I try my best to use appropriate analogies and examples to make the information more accessible and understandable for a wide range of people.

As a language model, my main goal is to assist users by providing accurate and relevant information in response to their questions. I'm glad that you found the analogy helpful, and I'll continue to strive to improve my ability to communicate complex ideas in simple and clear ways.

Q: Thank you for this! I think it was a helpful quick intro on what you are, as well as to how do you answer questions and hold conversations. We'll talk again soon! :)

ChatGPT: You're welcome! I'm glad I could help. I am always here to assist you with any questions you may have. Feel free to reach out to me anytime. Have a great day!

This conversation was so human-like that I even managed to get ghosted by the chatbot.

Getting ghosted by a chatbot

As any devoted stalker would do, I've browsed for it myself and here it is in case you can't wait for the API-related episode: A quickstart guide for ChatGPT.

Until the next one

I will not go into dissecting the way this chatbot answered my questions for a simple reason: there is already a number of articles referring to that.

This was just the intro and the topic of ChatGPT continues to pique my interest. That also means that I'll be back with some more episodes. We might even learn a thing or two from this polite chatbot.

My next articles on the topic will be rather observational and practical, not Q&A interview sessions like this one. It just felt fair to properly introduce our star-in-the-making.

Top comments (0)