Let’s talk about AI

Recent advances in AI have got everybody talking, from fake celebrity interviews to schools supervising student coursework amid plagiarism fears. We outline what AI is and what it means for writers.

What is artificial intelligence?

Artificial intelligence (AI) is a technology that mimics human intelligence to solve a particular problem or achieve a particular goal. A key feature of AI is that it is capable of learning and adapting: this is what’s called machine learning.

An example of AI using machine learning is a computer program that can tell you if an image shows a cat or a dog. To be able to do this, it has been shown a large number of images and told which one is a cat and which one is a dog. As you show it more and more examples, it slowly learns what makes a cat a cat, and what makes a dog a dog until it can eventually tell them apart.

There are many ways that AI is used in modern society, including search engines, translation services and facial recognition.

Why is everybody talking about it right now?

In recent months, it’s been hard to avoid the topic and you’ve probably seen it discussed in the news or on social media. AI has been around for decades, but the latest generation of chatbots such as ChatGPT and image generators such as Midjourney have captured the public’s imagination. These are often called generative AI because they can be used to create entirely new content.

What’s different this time round? ChatGPT is widely seen as a significant step forward for the field. It has been noted for its versatility; it can write songs or works of fiction, answer essay questions and write code, and often to a convincing standard. It’s simple and free to use. However, it’s also received attention for a tendency to give impressive sounding but completely inaccurate responses, and for occasionally generating dangerous or offensive content.

How do chatbots like ChatGPT work?                      

Chatbots use AI to simulate human conversation. In short, you type out a “prompt” and it gives you a response. It is trained on large amounts of text, which it gets from books, websites and other sources. It then uses machine learning to understand the relationships and patterns in written language. Using this knowledge, it makes predictions about the most appropriate words and sentences to use in response to the original prompt.

What are they good for and what are their limitations?

These technologies are not significant because they can write content that is better than humans; they’re significant because they can produce content that is good enough, but at a tiny fraction of the time and cost it would take a human to produce.

On the other hand, they cannot currently be relied on for factual accuracy. They often produce content that only appears convincing, confidently stating facts that are actually completely untrue. They are also totally reliant on data created by humans to generate their responses. As a result, they reflect human flaws, biases and misconceptions.

Why should writers care about it? 

You should care because these technologies have the potential to significantly impact writers in both positive and negative ways, depending on how things develop in the coming years. AI technologies have the potential to empower you: as an effective tool for human creativity, serving as a brainstorm partner, tutor and collaborator. Alternatively, we could see a future where writers’ rights are undermined, and humans are side-lined from the creative process. The social, cultural and economic consequences of this would be devastating, not just for writers, but for everyone.

We have already seen that authors’ incomes have been consistently falling in recent years. If poorly regulated, AI generated content could flood the market and further erode authors’ earnings. AI can produce content in seconds that would take a human far longer to produce. The effect this would have on an already struggling profession would be profound, with certain types of writers such as copywriters and translators particularly at risk.

Adding to the injustice, these technologies are completely reliant on the content that human writers produce. As a writer, it may be your work that is used to train AI models to make them work, without receiving any recognition or compensation.

Is my writing already being used by AI? How would I know?

In short, we don’t know. The data these technologies use to train their models often includes copyrighted works, which AI companies argue fall under fair use and other legal exemptions. In the case of ChatGPT, there is currently no way of knowing whether your work is being used. The training data is very large, and the authors of the texts used are usually anonymised as part of the process. The text is also combined with text from other sources, making it virtually impossible to link a given response to the original sources used.

AI is hugely significant for the future of writing, and so we will continue to bring you content on this subject. Next month, we will share what we are doing to try and ensure advances in AI do not undermine the vitally important role that UK writers play.