ChatGPT Unwrapped: A Beginner's Guide to Generative AI

Initially, I was hesitant to join this conversation, given the extensive coverage this topic has already received. Numerous headlines like "The Rise of ChatGPT: Are We Handing Over Control to AI and Losing Our Humanity?"; "Job Apocalypse: ChatGPT's Role in the AI-driven Unemployment Crisis", and "Battling ChatGPT-Assisted Cheating: Educators Struggle to Maintain Academic Integrity in the AI Age" have captured public attention.

However, my perspective changed when my 13-year-old daughter inquired about ChatGPT. She must have come across one of these headlines and was curious about how it could be utilised in relation to her homework.

I concede that my first explanation, which went along the lines of "ChatGPT is a helpful computer tool that uses generative AI, a kind of artificial intelligence, to create text that looks like it was written by a person. You give the interface an input and it generates an output. It learns to use words, grammar, and context by examining lots of written information with the assistance of machine learning, neural networks, and transformer-related methods, following years of training" may have come across as too technical, since she didn't inquire further about it.

I decided to write an article that would be easily understandable to individuals without technical backgrounds, including my daughter and her friends. Rather than focusing solely on ChatGPT, I opted to delve into the broader concept of Generative AI to provide a comprehensive understanding. While I have been a keen observer of Artificial Intelligence for years, my interest lies more in its practical applications rather than the mathematical and statistical underpinnings. Since the release of OpenAI's GPT-1 in 2018, I have been an armchair observer of its evolution. Understanding these advancements is essential in navigating the ever-changing digital landscape and avoiding sensationalised headlines.

What is Generative AI?

Generative AI (GenAi) is a subset of artificial intelligence that focuses on creating new, previously unseen outputs, such as text, images, or even music. Artificial intelligence is like a smart computer brain that can learn and think, helping us solve problems and complete tasks in ways like how humans do.

GenAi does this by learning from existing data using machine learning techniques, which involve training models to recognise patterns and make predictions or guesses based on data. Machine learning is a way for computers to learn new skills and improve themselves by studying patterns and examples, just like how we learn from experience.

It is like a digital assistant that can generate human-like output based on the input it receives. It uses deep learning, a branch of machine learning that uses neural networks—inspired by the human brain—to process information. Deep learning is a technique where computers mimic the human brain using neural networks. A neural network is a computer system inspired by the human brain, made up of many interconnected neurons that work together to learn and solve problems.

How Does Generative AI Work?

GenAi works by observing and analysing vast amounts of data. It uses complex mathematical models called neural networks, which are like a computer's brain, to identify patterns and relationships within the data in the text.

Some generative AI models like ChatGPT use a special tool called a "transformer" to learn how to understand and use words. They learn in two ways: with help (supervised learning), and by itself (unsupervised learning). When learning with help, they look at lots of sentences and questions with their answers. They learn what words should come next in a sentence and how to answer questions by seeing examples. At the same time, these models also learn by themselves. They look at words and sentences to find patterns and connections, even when nobody tells it what to look for. It is worth noting that errors may occur during the learning process. Generative AI models mitigate these errors by relying on feedback, learning from their mistakes, and continuously improving their performance.

Once the generative AI has learned from all the text data, it stores this knowledge in its neural network. When you give it a topic or a question, the AI uses the connections and patterns it has learned to predict or guess the best words or phrases that should come next, creating an answer. It starts by choosing words related to your input and then builds sentences or paragraphs based on context and the information it knows, much like how we think and write when responding to a question or crafting a story.

ChatGPT: A Prime Example of Generative AI

ChatGPT, developed by OpenAI, has captivated the tech community and beyond since the launch of GPT-3.5 in November 2022. The buzz around ChatGPT stems from various factors. One key aspect is the debut of an intuitive interface, comprising a straightforward browser and textbox, which has made it remarkably simple for the public to engage with this fascinating technology in a conversational manner.

Additionally, ChatGPT boasts a diverse array of applications, encompassing virtual assistants, content creation, translation services, and even programming assistance. Its prowess in generating coherent and contextually appropriate responses has piqued the curiosity of businesses, educators, and individuals alike, who are actively exploring ways to harness its potential for optimising processes and enriching user experiences.

It offers many benefits, such as sparking creativity, improving accessibility, reducing costs, and increasing efficiency. It also supports learning, education, and innovation across industries.

Stay tuned for more articles delving into how this ground-breaking technology has the potential to revolutionise several industries, including education, healthcare, application development, entertainment and more.

Generative AI models, while powerful, face several limitations that can impact the accuracy of their responses. These limitations include data quality, lack of context, difficulty with common sense, bias and language nuances. Furthermore, generative AI models cannot update their knowledge in real-time, as they rely on fixed datasets for training, which may lead to outdated or incorrect information. Additionally, these models can prioritise generating human-like text over accuracy or relevance, a phenomenon known as "hallucination". Notably, ChatGPT tends to hallucinate (make things up) with confidence. As a result, it is crucial to remain aware of these limitations while using generative AI and to consult additional sources for verification when necessary. Our jobs are safe for now.

How to use ChatGPT

Leveraging ChatGPT is a straightforward and user-friendly process. Follow these steps to begin:

1. Visit OpenAI's website: Navigate to OpenAI's official website (https://www.openai.com/) and search for information on ChatGPT. Stay up to date on the latest releases and versions, as OpenAI frequently introduces updates. ChatGPT 4, the latest (paid) version , offers improved language comprehension, more precise responses, and enhanced context awareness compared to ChatGPT 3.5, the default (free) model, due to advancements in architecture and training data.

2. Register or log in: If user authentication is required, sign up for an account or log in with your existing credentials.

3. Engage with ChatGPT: Upon accessing the platform, locate the text input box for entering your prompt. A ChatGPT prompt is an input text or question posed to the ChatGPT model, initiating a conversation, or seeking information, which the model responds to with contextually relevant and informative answers. The quality of the prompt directly influences the response quality. The significance of effective "prompting" will be discussed in the next article. Type your text and press "Enter" or click the "Submit" button.

ChatGPT will analyse your input and generate an appropriate response. It is important to bear in mind that ChatGPT possesses certain limitations, as has been previously stated. Therefore, in the event that doubts arise regarding the accuracy of responses provided, it is recommended to verify the information through additional means.

Until next time.

#generativeai #artificialintelliegence #chatgptai