AI – The Future of Writing?
Last month I was at a writing event at The Roundhouse in Camden. At the same time, The Roundhouse was also hosting an art exhibition called I Generate Bears, a series of collaboration pieces between a human and AI. This article is partly inspired by that exhibition (1) and recent events. I hope you enjoy.
Above is one of the creative art pieces from the I Generate Bears exhibition, created by Jess Thom and the AI program DALL-E 2.
The art of writing is always developing. First there were pictographs and hieroglyphics, then clay tablets, parchment and scrolls and eventually quills, pen and paper and even the keyboard. While the history of writing is fascinating (2) it also raises a very interesting question. What’s the next step?
The answer could be AI, Artificial Intelligence. The definition of what AI is rather tricky but to simplify, Artificial Intelligence is a computer or a computer programme that thinks or acts in a human like manner. Although you may consider AI to be bound to the realms of Science Fiction (think HAL from 2001: A Space Odyssey, Agent Smith from The Matrix or even the Terminator from the Terminator franchise) we use AI in our everyday lives. Whether you are composing a text with the help of autocorrect, scrolling through your social media feed or speaking to a virtual assistant such as Siri, Alexa or Cortana you are using AI.
Recently the American owned company, Open AI launched ChatGPT (Sorry, there will lots of acronyms in this post) a Chatbot that is trained to follow an instruction in a prompt and provide a detailed response. I asked ChatGPT the following:
This is what it came up with.
Not too bad, eh?
There are hundreds of Chatbots like ChatGPT but the reason this one is drawing attention in the mainstream media is that it is… almost convincing.
The future of AI is an inspiring yet also terrifying subject that is far too complicated for me to break down in a single web post. Instead, I’ll just be focusing on the writing aspect and I’ll attempt to answer the question, is AI the future of writing?
It is important to realise that AI isn’t a new phenomenon. Although the exact origins of AI are debated many experts claim that Alan Turing, the creator of the Turing Machine and the man who broke the Nazi’s Enigma Code, laid the groundwork for AI. The first ever AI programme was created by Christopher Strachey in 1951 (3). Strachey’s programme played checkers, learned from its human opponents and eventually outsmarted them. While this is impressive, the game of checkers is very different to writing a novel.
There have been previous attempts to teach AI to write. Most notably in 2017, a company called Botnik Studios wrote a new chapter of the Harry Potter series after creating a programme that had read all the previous books in the series.
Unfortunately, it seems the programme wasn’t fully able to grasp the intricacies of the English language nor the lingo of the Harry Potter universe itself. In 2017 Ross Goodwin drove from New York to New Orleans with an AI on his laptop connected to his car. This AI was able to write a novel based on the data it was fed. The novel was meant to be in the style of Jack Kerouac’s On The Road but the prose was choppy and nonsensical. Here is a quick extract:
“The time was one minute past midnight. But he was the only one who had to sit on his way back. The time was one minute after midnight and the wind was still standing on the counter and the little patch of straw was still still and the street was open” – the Road. AI
The difference between these examples and ChatGPT is that ChatGPT could pass as human if you were reading it without context. You could easily assume that the example of Gary the ballet dancer was written by a human, maybe a child with an impressive vocabulary but still a human.
While ChatGPT can generate highly convincing responses in many situations, it may also sometimes produce nonsensical or inappropriate responses, especially when presented with novel or complex tasks. Additionally, ChatGPT’s responses may be biased or reflect the biases present in its training data.
I didn’t write that last paragraph. ChatGPT did. You can probably tell as that isn’t my style of writing but maybe you didn’t realise that until the second or third line? When I was in Camden at The Roundhouse I struggled to tell which pictures were created by a human and which were AI creations.
While this technology is impressive it is also worrying. You would have no doubt seen the latest reports on AI in the news and online. At time of writing, Doctor Geoffrey Hinton has just resigned from his position at Google explaining that soon the AI Programmes will become smarter than their human operators. The news article itself can be found by clicking anywhere in this paragraph. Should we listen to these warnings or is it just doom mongering and exaggerations?
As ChatGPT itself has pointed out, it doesn’t understand complex tasks such as “write me a novel”. At the moment the programme is limited to about five hundred word answers. While you can use AI to generate writing prompts, character names and plot twists, It wouldn’t be able to write a great novel for you.
In addition to not understanding complex tasks, AI lacks two other key aspects that make us human. Emotions and creativity. Emotions connect us, we use our emotions to gauge how we are feeling and to recognise the moods of other people people. Our emotions motivate us, empower us and guide us through our lives. Even negative emotions such as sadness and anxiety are important so that we can appreciate our more powerful positive emotions against them. A robot just follows commands, it isn’t sentient, it doesn’t have dreams or ambitions. It just does whatever it’s told to do. Robots do. Humans feel. Similar statements can be said for our creativity. We use our creativity to solve problems, entertain ourselves and to improve our mental health.
I don’t intend to downplay the dangers of AI. I fully believe that it will become a global problem in the future but I don’t think Skynet is here just yet. But it’s close. While AI may struggle to replace writers and human creativity I do fear for other aspects of humanity. It will be interesting to return to this post in five or ten years time to see what the latest advancement in this technology is.
Thank you all for reading. I’ll see you next time when I will, hopefully, write about a more cheerful topic.
Footnotes
(2) And worthy of an article itself in the future perhaps?
(3)Interestingly the first example of AI dates back to an Ancient Greek myth. Talos was a giant automaton forged by the god Hephaestus to protect the island of Crete. He would patrol the island three times a day and throw boulders at approaching ships. He was defeated by Jason and the Argonauts who simply unplugged him… I thought this was a piece of Wikipedia vandalism when I first read that but no, Talos was a real character in Greek myology.