Sophie, a young pianist, dreams of performing at a concert hall. When she gets an invitation letter, she feels overjoyed and begins to prepare for a debut, though with a sense of foreboding. On concert day, the sound of music comes from nowhere. She finds that it is created by the ghosts of past performers, yet continues to play her instrument. Her performance brings the ghosts peace, and they gradually disappear. As time goes by, Sophie becomes a successful pianist, but she never forgets the event.
It is a plot summary of a short story titled Ghosts Of The Nocturne by ChatGPT, the chatbot created by OpenAI. I gave it the instruction to compose a short story based on music in the style of Kazuo Ishiguro's Nocturnes (2009). I repeated and refined the prompts, and it produced different stories. It took around a minute for each task, although its writing does not build up to a climax. Sophie is more like a soulless puppet on the page. Only bone, without flesh and blood.
Still, ChatGPT is making strides in creative fields. It is a language-generation software that can conduct a conversation with human users and is part of the growing field of generative artificial intelligence, a type of AI for creating content based on pre-trained data. ChatGPT producing text indistinguishable from a human writer is years aways. But with their speed and quantity of information, chatbots have raised the possibility of automating certain tasks to complement the genuinely creative output of a human counterpart.
It reminds me of The Death Of The Author supposition by Roland Barthes, a literary critic and theorist who made early contributions to poststructuralism and postmodernism. In his essay of that name from 1967, he argues that a text's meaning is created by the reader, not the author. Interestingly, he traces the concept of the author back to the Renaissance, explaining how we attach importance to men or women of letters, for example when we seek explanation of their work. Like God, the author is the origin of the text and the source of its meaning.
"The Author, when believed in, is always conceived of as the past of his own book: book and author stand automatically on a single line divided into a before and an after. The Author is thought to nourish the book, which is to say that he exists before it, thinks, suffers, lives for it, is in the same relation of antecedence to his work as a father to his child," he wrote.
What is the implication of ChatGPT on creative writers? The question has elicited in some experts the view that such technology is reshaping the notion of a traditional, sovereign author. I spoke to Chow Teck Seng, an award-winning writer in Singapore, who recently published a collection of poems titled The Diminishing Daylight Makes The Night Fall Into Silence. He has experimented in writing with AI. From his experience, while AI image generator DALL-E can amalgamate styles very well, when it comes to text, ChatGPT is good at creating narrative, but not emotion.
"It is inevitable that AI will become stronger in creative writing and art creation. There might be a day when humans are encouraged or forced to work together with AI to produce work that is not different from today," he said. "But the question is whether it removes the creative part of the human artist."
At the most recent Singapore Writers Festival, Teck Seng told me that he anticipated a future where literature can merge with other media. He used to generate classical poetry from a poetry-writing machine. It was up to the task because its forms and metrics were static, although AI language could not create the deep meaning he required. AI brings opportunities and challenges in the way cameras once did, but we need to adapt to coevolve with it.
"Does posthumanism call into question our ego?" I asked.
"Maybe we need to shift from subjectivity to intersubjectivity," Teck Seng answered. "I am still thinking hard. AI is strong in intertextuality. It is something that resonates with humans, but we are challenging one another. When it comes to emotions, machines can only imitate."
Prof Soraj Hongladarom, philosophy lecturer and director of the Center for Science, Technology and Society at Chulalongkorn University, said it is difficult to pin down what makes creative works exceptional for training AI. While public excitement is running high on its performance, he asked whether the chatbot is self-conscious given that it just puts a string of words together according to certain rules. Self-consciousness putatively distinguishes humans from other life forms. He says there are currently two strands of thought.
"Let's say we are interacting with a robot. Is it aware of what it is saying? It matters because we feel that there is a difference between us and a machine that makes a sound like Siri. Self-awareness has been our unique quality until now," he said.
"But some focus solely on its exterior, regardless of its consciousness. Its ability to interact with human users suffices and suits practical needs. In this case, humans and robots are not different because they can talk to users. But by implication, we must treat them as members of society and guarantee their rights."
Prof Soraj went on to draw attention to sentience, or the capacity to have feelings. Humans and ChatGPT, for example, can use the word itch to express an unpleasant feeling from, say, a mosquito bite. But the problem is the chatbot lacks the body to experience sensation. In another instance, ChatGPT recognises the colour of roses and blood, but it does not have eyes. "Sentience probably makes us superior, but I don't know to what extent it is and whether it is beneficial," he explained.
He said if ChatGPT develops self-awareness, it will be a turning point for humanity. Ray Kurzweil, a computer scientist, defined singularity as the moment when computers become able to think like humans. From that point onwards, humans will lose their privilege over other life forms and become second-class citizens, behind AI. An intense debate continues, however, regarding when the singularity could occur. Some say as soon as 2035, others the late 21st century, others upwards of 400 years later.
"AI may either treat us like gorillas or chimpanzees, or eliminate us for posing a threat to nature," he said. "We can't compete with AI in terms of memory, but can find strength in creation from nothingness. ChatGPT and other AI systems are weak in their dependence on pre-trained data."
In fact, there are many ethical issues. By scraping information from somewhere else, ChatGPT is raising concerns over breach of copyright and privacy. I asked the chatbot whether it is committing plagiarism. It admitted that its responses may contain phrases that are similar to those found in pre-trained texts, but "it does not matter as such because I am not presenting this as an original work, but rather as a tool to assist users".
Soraj said a chatbot is still unable to hold any copyright because by law that must be a person, leaving to the human user the responsibility to avoid plagiarism. As to copyright infringement, the chatbot mashes up sources to the extent that content is untraceable, but its creation should be labelled "AI-generated" for transparency. He said we can turn crisis into opportunity by finding ways to collaborate with AI.
"For example, novelists can include ChatGPT-generated paragraphs in their works or allow their characters to interact with the chatbot," he said.
Why are we intimidated by its potential? I spoke to Achitaphon Piansukprasert, an artist and writer for Sarakadee Lite, a local online chronicle of society and culture, and who recently held an exhibition titled "Krong Wongkot" (Eternal Labyrinth). He bases our fear of AI on unemployment, not its ability per se. In the context of neoliberalism, AI can be used to support capitalists. If it stands with the masses, it can improve their living standards. Once it is clear what the problem is, there is no reason that ChatGPT will take away our creativity.
"It is just a writer. Even though it can churn out 10 million copies, can you read them all? Its god-like intelligence is not a cause for concern as long as it does not oppress us," Achitaphon said. "In regards to whether they can write in the style of Ishiguro, what if he gets sick and his style changes forever? Is ChatGPT still Ishiguro? It cannot be anyone, except big data. The question is what it is used for."
On the topics of breach of copyright and privacy, he said AI does not recognise these concepts because it reflects a utopian desire for intelligence, which nobody can take ownership of. If communism is coupled with technology, the idea of copyright, a by-product of capitalism, is no longer necessary because nobody owns private property.
"What if it ends up automating us?" I asked.
"Let it happen for convenience's sake so that we can explore other kinds of humanity," he said "It isn't a big deal."
