Less than two months ago, at the end of November, developers at OpenAI released ChatGPT for use online. It’s currently free to use (though there are plans to monetise it), and in just under four days, it had over a million users. But what’s the fuss all about? Some are considering ChatGPT as ‘just another Google,’ yet this could not be further from the truth. Unlike Google, ChatGPT gives quick and detailed responses in a human-like way to a wide variety of questions and requests; from writing cover letters, to fictitious plays, to academic articles. The dizzying potential of ChatGPT has intrigued - and agitated - many.

But what is ChatGPT exactly? I suppose the best thing to do is to let it explain in its own words. The following are ChatGPT’s answers after asking it to give short and concise replies for a newspaper article.

Can you explain, as simply as you can, what you are?

ChatGPT - I am a computer program called a language model. My purpose is to understand and generate human language. I am trained on a large dataset of text, allowing me to understand and respond to text input.

So how do you go about answering the questions posed to you?

ChatGPT - I use a technique called machine learning to generate responses to questions. This involves analyzing patterns in large amounts of text data, and using that information to generate new text that is similar to the text I was trained on. When I receive a question, I use the patterns I have learned to generate a response that is relevant to the question.

In effect ChatGPT uses probability to guess what a suitable reply to a prompt should be, by comparing it to similar data from the Internet, and learning in the process (this also explains why, at times, its replies are factually incorrect). This is not really what we understand by ‘thinking,’ and phrases such as ChatGPT ‘having a mind of its own’ should be taken lightly.

Thinking or not, however, the fact still remains that ChatGPT has great potential, from uses in education, business, marketing, and law, amongst others. Though, at the same time, many are seeing ChatGPT as another example of how AI and automation will make large amounts of people redundant. Here’s what ChatGPT had to say about this:

Do you think you will make a lot of people lose their jobs?

ChatGPT - Language models like me have the ability to automate certain tasks that were previously done by humans, such as answering customer service inquiries or writing news articles. It's possible that my technology could lead to some job loss in certain fields, however it could also lead to new opportunities and efficiencies in other areas. Additionally, my technology can be used to augment human capabilities and improve productivity, making it a complement to human labor rather than a replacement.

Easier said than done. As evidenced by a recent newspaper article, so far the worry that has been expressed in the local education sector is limited to ChatGPT being used to help students cheat in assessments. Yet this is only one aspect of a multitude of implications ChatGPT (and its successors) will bring. What we are in need of now, therefore, is a paradigm shift in our education system - something which (apart from the introduction of coding to younger students) we have not yet begun to consider.

There are also wider, and possibly more sinister dangers that more sophisticated systems like ChatGPT can pose. Fake news articles, malware and phishing emails can be created in seconds. More unsettling is the fact that - similar to other AI programs - it seems to exhibit biases and a lack of basic moral principles. And while its developers have tried to install some safety features to curtail the above, these measures have so far proved ineffective. So, naturally, I asked ChatGPT:

Do you think you're a danger to society?

ChatGPT - As a language model, I am a tool that can be used for a variety of purposes, some of which may be beneficial and some of which may be harmful. It is ultimately up to the users of my technology to determine how it is used, and for what purposes. It's important to use the technology ethically and responsibly.

So it seems AI can’t answer all our questions after all - if anything, it might be creating some new ones! Some might argue that ethicists and philosophers might be over-panicking. And yet, a study published last year found that only 1% of the most highly cited machine learning papers discuss the negative potential of AI. The need, then, for ethical reflection is real and urgent.

Delivering his speech for the inauguration of the Centre for the Future of Intelligence in Cambridge, Stephen Hawking said, “The rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity. We do not yet know which.” Indeed, as time passes, and more revolutionary innovations are made, the potential benefits that can be reaped rise exponentially - but so do the dangers. If Hawking’s prediction is to be heeded, the rapid technological advancements we are seeing need to be accompanied by a serious interdisciplinary effort amongst scientists, philosophers, developers and policy-makers alike to understand the implications, and guide the process - something which ChatGPT can’t do for us just yet.

 

Fr Jean Gové is reading for a doctorate in philosophy of mind at the University of St Andrews, Scotland.