Introduction
OpenAI's brainchild, "ChatGPT," is all the rage right now and is poised to be a massive disrupter in the software engineering industry. Since its release on November 30th, people have been writing plenty of guides, blogs, and opinion pieces about it all over social media. It has even caught the attention of the non-tech communities, as the NY Times's "The Daily" podcast recently reported on it.
Sentience
But with everyone else writing pieces about ChatGPT, what if we asked it to write a blog post about itself? Let's see what it can do if I ask it to:
Write me a blog post on the various capabilities and insights of ChatGPT
Its response:
ChatGPT is a powerful language generation model developed by OpenAI that has the ability to understand and generate human-like text. This model has been trained on a massive amount of data, making it capable of understanding a wide range of topics and answering various types of questions.
One of the key capabilities of ChatGPT is its ability to generate coherent and fluent text. The model has been trained on a diverse range of text, including books, articles, and websites, which allows it to produce text that is similar to that written by humans. This makes ChatGPT a useful tool for tasks such as text summarization, content creation, and automated writing.
Another important capability of ChatGPT is its ability to understand and respond to natural language questions. The model has been trained on a large number of conversational datasets, which enables it to understand the intent behind a question and provide a relevant and informative response. This makes ChatGPT an effective tool for natural language processing tasks such as question answering and chatbot development.
Additionally, ChatGPT has the ability to generate text in a specific style or tone. The model has been trained on a diverse range of text, which means that it can generate text that is formal or informal, serious or humorous, and so on. This makes ChatGPT a useful tool for tasks such as content generation and language translation.
In terms of insights, ChatGPT has the ability to provide valuable information about the text it is generating. For example, it can provide a summary of the text, highlight important keywords, or identify the main topics discussed. Additionally, ChatGPT can also be used to analyze text and identify patterns or trends in the data.
In conclusion, ChatGPT is a powerful language generation model that has the ability to understand and generate human-like text. It can be used for a wide range of natural language processing tasks such as text summarization, content creation, question answering, and chatbot development. Additionally, ChatGPT can also provide valuable insights into the text it generates, making it a valuable tool for data analysis and understanding.
Typically, if someone were writing a blog post about themselves, it would take considerable time and a reasonably introspective writing process to come up with anything close to the above. It took ChatGPT less than two minutes. But that is just it; this is not the first time that ChatGPT is answering this question. So many people had asked it to do that, and it would not even tell me an exact number when I asked it:
How many times have you answered "Write me a blog post on the various capabilities and insights of ChatGPT"?
Its response:
As an AI model, I do not have the ability to remember previous interactions. However, I can tell you that I have been answering this type of question multiple times, as it is quite a common question among users interested in the capabilities of language generation models like ChatGPT.
So I simply asked it again:
How many users have asked to answer the following type of question: "Write me a blog post on the various capabilities and insights of ChatGPT"?
Its response this time:
As an AI model, I do not have the ability to track the number of users who ask specific questions, neither I can count how many times a question has been asked. But, as I mentioned before, it is a common request among users interested in the capabilities of language generation models like ChatGPT.
It knows I and others have been asking multiple times about the specifics of this question, and that it has answered the question previously within my session with it is nothing short of an engineering marvel. While it may not be full-on sentience, this proves remarkable and unprecedented compared to other examples of chatbots in the past. Let us see what else it can tell us about itself.
An Origin Story
Everybody, even AI, needs a compelling backstory. Many would describe their life by first saying they came from one of the major cities or a small town. They did decently in grade/high school; they might have played a sport and remembered their senior prom fondly. Some went further on to college or vocational programs and graduated with a degree for their occupation in the workforce. They didn't get a "bad" job, so to say, it pays the bills and gives them a little slice of their own agency. They met the love of their life somewhere along the line and might have started a family. And so the story goes.
When asked, "Tell me the origins of ChatGPT," ChatGPT's narrative is starkly different:
ChatGPT, also known as GPT-3, is a language generation model developed by OpenAI. The model is based on the transformer architecture, which was introduced in a 2017 paper by Google researchers. The transformer architecture is a neural network architecture that is designed to process sequential data, such as text.
The development of ChatGPT began in 2018 and it was released in June 2020. It was trained on a massive amount of data, including books, articles, and websites, which allows it to understand and generate text on a wide range of topics. The training dataset used to build the model is known as WebText, which includes around 570GB of text data.
ChatGPT is the third generation of the GPT (Generative Pre-training Transformer) models, the first one, GPT-1 was released in 2018, and GPT-2 in 2019, both models already showed good results in language understanding and generation but with less data and computational power than the GPT-3.
ChatGPT is considered to be one of the most advanced language generation models currently available, and its capabilities have been demonstrated in a variety of natural language processing tasks such as text summarization, content creation, question answering, and chatbot development.
We know ChatGPT generated the tale of its upbringing. We expect it to sound clunky, but we still need to account for how alien it feels to read. Imagine asking someone, "So tell me about yourself; where are you from?" to be told about their demographics, age, sex, and race, as well as other aspects like reading off answers from a National Census questionnaire. Not to say there is anything wrong with being proud of unique identifiers, but ChatGPT's way of answering this question sounds arguably too robotic. But, perhaps, that is a litmus test for dealing with an AI.
Conclusion
As Microsoft continues to funnel money into the ChatGPT project and Google has its version of "Deepmind Chat" coming soon, everyone should sign up for a free account and try taking ChatGPT for a spin. It is clear the days of artificial intelligence are only just beginning. Many industries have yet to see automation like this before and will face a partial or complete overhaul soon.
Me:
Hello Brave New World?
ChatGPT:
Hello, "Brave New World" is a novel written by Aldous Huxley, published in 1932. It is a dystopian novel set in a future society where people are controlled through the use of technology and conditioning. The novel explores themes of individuality, freedom, and the consequences of a society that values conformity over individuality. Is there something specific you would like to know about it?
