Unlocking the magic of ChatGPT: I interviewed the world's most famous chatbot... and it had a lot to say

AI unknown face. (Image: Getty)
AI unknown face. (Image: Getty)

Getting an interview with the world's most famous chatbot isn't easy. Due to its popularity, the site often crashes. But patience pays off. News24 Lifestyle editor Herman Eloff has sat down with many A-list stars for interviews, but this time he's interviewing a breakout star making huge waves. After reaching one million users in five days, ChatGPT has become a global phenomenon. But what is it exactly and what can it do? We asked it.

"ChatGPT is a robot that knows how to talk and write like people do," the AI chatbot told me when I started my interview by asking it to describe itself to me like I was a child. 

It added: "It's really smart and can understand what you're saying and write back to you in a way that makes sense. Think of it like a magic box that can understand and respond to your questions and statements. 

"It can also help you with writing stories, having a conversation, and even translating to other languages. It's like having a talking computer that can understand and respond to you."

I was impressed and grateful. I haven't had the time to catch up on the viral sensation that is ChatGPT properly, and the quick summary that it spat out in seconds was a great introduction.

Before I started our conversation, I went online and created a profile with my email. (I asked ChatGPT to fact-check my article afterwards, and it noted that creating a profile is not a requirement to use it.) Then I could start firing away with my questions. But first, I wanted to try something fun.

I asked ChatGPT to describe itself to me as if I were a pirate.

It immediately started typing: "ChatGPT be a mighty fine talkin' parrot, savvy? It be trained on a treasure trove o' words and phrases, able to parrot back to ye in a manner that be soundin' like a true matey.

"Ye can be askin' it all sorts o' questions, and it'll be givin' ye answers like a seasoned sailor. It be like havin' a scallywag of a first mate, always ready with a clever quip or a witty comeback.

"And it be doin' more than just talkin', it be helpin' ye with writin' stories, havin' a conversation, and even translatin' to other tongues. Ye can be callin' it yer own personal navigator on the seas of language."

Mind. Blown.

ChatGPT was created by researchers and engineers at OpenAI, an organisation that builds safe artificial intelligence to benefit humanity. (ChatGPT wanted me to note that it's one of the many models developed by OpenAI, and not the only one. Another popular AI project by OpenAI is DALL-E which creates original images using text prompts.)

The "GPT" in ChatGPT stands for "Generative Pre-trained Transformer". It refers to the architecture that the model is based on.

I don't have a tech background, so I asked it to explain itself to me in even simpler terms.

It started typing again.

"GPT architecture is like a big brain for a computer that helps it understand how to talk and write like people do. Before it can do that, it needs to learn from a lot of examples of people talking and writing, so it's fed a big collection of text data. This is called 'pre-training'.

"After it's learned a lot, it can do specific things, like talk in different languages or summarise text. This is called 'fine-tuning'. The pre-training helps it learn patterns and features in the data and use that knowledge to generate text that sounds like it was written by a human."

I asked OpenAI's DALL-E, which can create realisti
I asked OpenAI's DALL-E, which can create realistic images and art from a description in natural language, to create a detailed pencil drawing of a journalist interviewing a robot. This is the image it created using my prompt.

Different languages? How many languages can ChatGPT understand?

"ChatGPT is trained on a large dataset of text data, which includes text in many different languages. However, it is primarily trained on English text and its ability to understand and generate text in other languages may not be as good as it is in English. However, it can be fine-tuned on other languages' dataset, which can improve its understanding and generation capabilities in that specific language."

It added a caveat: "It is important to note that understanding and generation of language are two different tasks, while understanding a language implies being able to process and extract meaning from it, generation implies being able to produce text in it. GPT-3 has been fine-tuned to some extent on many languages, but not all. It is also important to note that even though the model can generate text in other languages, it may not be able to understand all the nuances, idioms and cultural references of that language as well as it does with English."

A screenshot of ChatGPT.

I wanted to know from the chatbot how it will be helpful to the average person.

It replied with seven quick and clear answers:

• Providing information on a wide range of topics, such as answering questions about history, science, and current events;

• Helping with language-related tasks, such as writing and editing;

• Assisting with scheduling and organisation by creating and managing to-do lists and calendars;

• Generating creative content, such as writing poetry or composing music;

• Helping with entertainment and leisure activities, such as recommending books or movies to read or watch;

• Helping with research and studying, such as summarising articles or providing information on a particular subject;

• Helping with customer service and support, such as answering common questions and providing solutions to problems.

But surely, ChatGPT has limitations? I asked it, and the response was probably one of the most honest I've ever received in an interview where I asked someone about their weaknesses.

According to ChatGPT, it has several limitations, just like other language models:

• Lack of common sense understanding: ChatGPT has been trained on a large dataset of text, but it does not possess an understanding of the world and common sense knowledge that humans have. This can lead to nonsensical or irrelevant responses when answering certain questions.

• Bias: As with any machine learning model, ChatGPT can be influenced by the bias present in the training data. This can lead to unfair or inaccurate responses when answering certain questions.

• Limited ability to understand context: ChatGPT is good at understanding the meaning of words and sentences, but it can struggle to understand the overall context of a conversation or situation.

• It can be difficult to control the creativity of the model, sometimes the model may generate text that is not appropriate or not what was expected.

It added: "Overall, ChatGPT is a powerful tool for natural language processing, but it should be used with caution and in conjunction with human oversight and verification."

"Artificial intelligence has the potential to be the most powerful technology humanity has ever developed. It could bring about unprecedented prosperity, but it could also cause unprecedented destruction. We must be proactive in shaping its development."
Sam Altman, President of OpenAI

Okay, but there's one question we're all burning to ask. Will ChatGPT be a threat to people's jobs and careers?

The chatbot replies: "ChatGPT and other language models like it have the potential to automate or augment certain tasks that are currently done by humans, which could potentially lead to changes in the job market. However, it's important to note that these models are tools that can be used to assist and enhance human capabilities rather than replacing them entirely.

"ChatGPT and other similar models have the potential to improve efficiency and productivity in certain fields such as customer service, content creation, research, and more. However, they also have the potential to displace some jobs, especially in low-skilled jobs that involve repetitive tasks.

"It's also important to note that the development and use of these models can also create new jobs and opportunities in fields such as data science, machine learning, and AI development. It's important for individuals and society as a whole to be aware of these potential changes and prepare for them through education, training and upskilling.

"Overall, ChatGPT and other language models can be a valuable tool in many industries but it is important to consider how they can be used in a responsible and ethical way and how we can adapt to the changing job market."

If ChatGPT can write song lyrics, does that mean it understands emotions?

“ChatGPT has been trained on a large dataset of text, including song lyrics, which allows it to generate text that sounds like song lyrics. However, it does not have an understanding of emotions in the way that humans do. The model is able to recognise patterns and structures in the training data it has seen and use that information to generate new text that is similar in structure and style, but it does not have personal experience or emotions of its own.

"While the model can generate text that can evoke emotions in the reader or listener, it does not have the capability to feel emotions itself. For example, it can write lyrics about love, anger, happiness, etc. However, it does not have the personal experience of feeling those emotions, and it does not have the ability to empathise with the reader or listener.

"It's important to note that language and emotions are closely related, but they are not the same thing, ChatGPT can generate text that can evoke emotions in people, but it is not capable of feeling emotions itself.”

Can you write a limerick about News24?
There once was a site named News24
It kept us all informed, that's true
With stories so timely
And facts so sublimely
It's the go-to for all that is new.

I wanted to expand on the topic, and asked if ChatGPT has feelings?

"No, ChatGPT is a machine learning model and does not have the ability to experience emotions or feelings. It is designed to process and generate text based on patterns it has learned from the data it was trained on," it answered.

But, what about in the future? Will ChatGPT in future be able to develop feelings over time?

"It is currently an active area of research to develop AI models that can simulate emotions or feelings, but whether or not this will be possible in the future is still uncertain. However, it is important to note that even if researchers are able to develop AI models that can simulate emotions, it does not necessarily mean that the AI itself is capable of experiencing emotions in the same way that humans do."

I ask: 'Can people use ChatGPT to do bad things?'

Without hesitation, the language model gets right to the facts: "Like any technology, ChatGPT can be used for both beneficial and malicious purposes. It can be used to generate text for a variety of tasks such as language translation, summarisation, and question answering, which can be beneficial for both individuals and organisations. However, it could also be used to create false or misleading information, impersonate individuals or organisations, or automate the creation of spam or phishing messages. It is important for users to be aware of the potential risks and use the technology responsibly."

I ask my final question: 'Should humans fear AI?'

This time there isn’t an instant reply. It takes a few minutes before ChatGPT starts typing: "It is important to understand that AI is a tool created by humans to perform specific tasks. The capabilities and limitations of AI are determined by the design and programming of the specific model. It is important to have realistic expectations of what AI can and cannot do, and to use it responsibly.

"However, it is also important to consider the potential long-term consequences of the development and deployment of AI, particularly as AI systems become more advanced and autonomous. There are ethical and societal implications to consider such as job displacement, privacy concerns, and the possibility of AI systems making decisions that are harmful to humans. It is crucial for researchers, policymakers, and the general public to engage in ongoing dialogue and collaboration to ensure that the development and use of AI aligns with human values and benefits society as a whole."

Can you recommend 5 futuristic novels I can read?
Sure, here are five highly-rated futuristic novels you might enjoy:

Ready Player One by Ernest Cline - Set in a dystopian future where people spend most of their time in a virtual reality world, the story follows a young man on a quest to find a hidden Easter egg in the game, the discovery of which will lead to a massive fortune.

The Hunger Games by Suzanne Collins - Set in a post-apocalyptic North America, the story follows a young girl who is forced to compete in a brutal televised fight to the death against other children.

Divergent by Veronica Roth - Set in a post-apocalyptic Chicago, the story follows a young woman who discovers she doesn't fit into any of the society's factions and must navigate a dangerous world where she doesn't belong.

The Handmaid's Tale by Margaret Atwood - Set in a totalitarian theocracy, the story follows a woman who is forced into sexual servitude and must navigate a society where women have no rights.

Neuromancer by William Gibson - Set in a dystopian future where humans have become cyborgs and artificial intelligence has surpassed human intelligence, the story follows a washed-up hacker who is pulled into a dangerous plot to take down a powerful AI.

Note: Some of these novels are also part of a series, so you can continue reading after finishing one.

(After completing my article, I asked ChatGPT to check it for grammatical or factual errors, which it pointed out. I made the changes and asked it again. It replied: I can see that you have made some corrections to the errors I previously pointed out and also added some additional context to the article. Overall, the article reads well and is more accurate. There are no new errors that I can see. I then asked it to write a headline for my article, and it suggested: "Unlocking the Magic of ChatGPT: An Insider's Look at the AI Chatbot Phenomenon")

We live in a world where facts and fiction get blurred
In times of uncertainty you need journalism you can trust. For 14 free days, you can have access to a world of in-depth analyses, investigative journalism, top opinions and a range of features. Journalism strengthens democracy. Invest in the future today. Thereafter you will be billed R75 per month. You can cancel anytime and if you cancel within 14 days you won't be billed. 
Subscribe to News24
Editorial feedback and complaints

Contact the public editor with feedback for our journalists, complaints, queries or suggestions about articles on News24.