Introduction.

If you’ve heard about the transformative power of “ChatGPT” but are wondering what exactly this AI tool is…. you’re not alone.

A quick scan through recent news articles gives some idea of the excitement and chatter surrounding this technology. From discussions about “game-changing” technology to rival products (such as Google’s “Bard” chatbot) and the legal implications of businesses using AI copy… it’s a brave new digital world.

In light of all this hype, ChatGPT reached 100 million users just two months after launch. This unprecedented take-up made the AI chatbot the fastest-growing internet app ever.

So, what’s all the fuss about? Let’s take a look.

What is ChatGPT?

ChatGPT is essentially a very clever and conversational chatbot.

Launched by the San Francisco-based company OpenAI in November 2022 – it’s a (currently) free service with some paid options.

You might have used similar technology in the past… and chances are, you might not have been impressed. Chatbots have historically produced pretty robotic phrases and sentences, often with humorous and downright nonsensical results.

But ChatGPT feels different.

It can easily create complete college-level essays (in fact, there are increasing concerns about students using this tech) as well as fictional stories, poems and job applications. It does this by drawing on a massive amount of internet data, combined with careful input from human experts.

Talking about the technology, Professor Michael Wooldridge (Director of “Foundational AI Research” at the Alan Turing Institute in London) describes it as the first system he could “genuinely get excited about”. Indeed, he stressed how it would take “1,000 human lifetimes to read the amount of text the system was trained on… hidden away in all of that text is an awful lot of knowledge about the world.”

So to find out more, who better to ask than ChatGPT itself?

We asked ChatGPT to describe itself in under 300 words.

Here are the results:

ChatGPT is a large language model developed by OpenAI that’s capable of processing natural language text and generating human-like responses.

The name “GPT” stands for “Generative Pre-trained Transformer,” which refers to the type of neural network architecture used to build the model.

ChatGPT is trained on massive amounts of text data from a diverse range of sources, including books, articles, and websites, using a process called “unsupervised learning”. During training, the model learns to identify patterns and relationships in the text data and use them to generate new text that is grammatically correct and semantically meaningful.

One of the key strengths of ChatGPT is its ability to generate natural-sounding language that closely resembles human speech. This is achieved through the use of techniques such as “attention mechanisms”, which allow the model to focus on specific parts of the input text when generating output.

ChatGPT can be used for a wide range of natural language processing tasks, including language translation, question-answering, text summarization, and chatbot development. It has been used in a variety of applications, from virtual assistants to customer service chatbots to language learning tools.

However, it’s important to note that ChatGPT is not a perfect solution and has limitations. For example, it may struggle with understanding context or generating responses that are consistent with a given persona. Nonetheless, it remains a powerful tool for natural language processing and has the potential to revolutionise the way we interact with machines.

How does ChatGPT work?

If phrases like “generative pre-trained transformer” and “neural network architecture” leave you feeling baffled – don’t worry! This is a perfect example of why a mix of AI and human editing is needed, to ensure answers are tailored to specific contexts.

In simple terms.

ChatGPT is a type of artificial intelligence (AI) system that uses machine learning to understand and generate human-like language. It does this with a “transformer” network – which is just a computing system modelled after the human brain. Pretty neat, right?

When you ask a question or make a statement to ChatGPT, the system breaks your text down into individual words and analyses the relationships between them. It uses a massive dataset of human text (called a “corpus”) to learn specific phrases and their use in context. This includes words and phrases from a wide range of sources, such as books, websites, and social media.

Once ChatGPT has analysed your input.

It generates a response by predicting the most likely phrases that would form part of an answer. This is presented in natural language, like it’s coming from a human user.

As more and more people use ChatGPT and provide feedback, the more the system learns and improves over time. This is called “training” the system, and helps improve the accuracy and relevance of its responses.

What are the potential uses of ChatGPT?

For ChatGPT’s creators, the technology aims to help users find information – providing accurate and helpful answers on all manner of topics. It lets people ask questions, seek knowledge and engage in discussions about pretty much anything.

Potential uses include:

• Educational settings: empowering school and university students to discover educational resources and create their own personalised learning experiences. It could even help with homework assignments and class essays.
• Language learning: helping students learn new languages by answering questions, providing vocab and grammar suggestions and conversation practice.
• Customer service: providing 24/7 support, dealing with frequently asked questions and helping with common issues.
• Travel: research new cities or areas with a single question, as well as help with booking flights, finding deals and suitable accommodation.
• Copywriting: are you entirely sure there’s a human behind this article? Hopefully. But ChatGPT has immense potential to transform copywriting and digital marketing services.
• Creative writing: this raises ethical concerns about intellectual property and an author’s own work – but ChatGPT could help with drafts of short stories, poems and even novels.
• Technical advice: from writing website code to healthcare queries and financial advice – ChatGPT can help all sorts of people and professions access the information they need.

In short, ChatGPT has the potential for use in pretty much any context where natural language is used.

What are some of the concerns about ChatGPT?

Like any new technology, there are several concerns and questions about the use of ChatGPT.

The company openly lists the tool’s limitations in their introductory blog post, citing things like “plausible-sounding but incorrect or nonsensical answers” and “excessively verbose” responses with an inability to ask clarifying questions.

Other potential issues include:

• Bias and fairness: as ChatGPT learns through massive amounts of internet discussion and data, it can perpetuate and reinforce any existing biases and inequalities in society.
• Privacy and security: ChatGPT could be used to collect personal data from users, including their questions and the answers given.
• Misinformation and fake news: ChatGPT could be used to spread misinformation or fake news, particularly if the training data includes inaccurate or misleading information.
• Lack of accountability: especially if ChatGPT is used in healthcare or finance settings (for example), how do we assign responsibility if information is incorrect? In turn, could we hold organisations accountable for any harm caused?
• Dependency: this is tricky to assess, but could users (particularly in school and educational settings) become too dependent on ChatGPT rather than their own research skills? This could potentially lead to declining critical thinking, essay-writing and independent decision-making skills.
• Unpredictability: with greater machine learning comes increased risk. It’s a little researched and understood topic, but could the technology interact with other systems or be used in subversive ways by rogue states and users?

 Are ChatGPT responses plagiarism free?

This is one of the biggest concerns about ChatGPT, so it’s worth covering this question in more detail.

Indeed, lecturers at UK universities have recently been urged to   review their course assessments, amid evidence students are already using ChatGPT for essays. While some lecturers, such as University of Wisconsin-Madison Philosophy Professor Lawrence Shapiro aren’t so concerned (despite reporting ChatGPT has his colleagues “shaking in their Birkenstocks”) – it’s a hot topic.

So, given its potential use (and misuse) by school and university students, what are the issues around plagiarism?

Well, remember ChatGPT responses are generated by machine learning that processes and analyses vast amounts of data. This means that while responses aren’t plagiarised in the traditional sense, they may contain phrases, sentences or even paragraphs that are similar or identical to existing text.

According to ChatGPT itself.

Responses aren’t “intended to be used as a substitute for original thought or creative writing.” The chatbot goes on to give students further advice, that if “using information from ChatGPT in your own work, it’s essential you properly attribute any quotes or ideas to their original sources.”

Feeling curious?

Sign up for a free account and start exploring ChatGPT’s functionality for yourself. Whether you’re up for a five-minute conversation, delving into the depths of philosophical conundrums or just some research help… it’s a useful and fascinating tool. The exact implications for intellectual property, privacy, accountability and plagiarism are still emerging though – so well and truly watch this space.