While writing my novel The Hummingbird Effect, I logged in to ChatGPT and asked the question: How would you destroy the world?
Programmed not to ‘provide guidance on how to cause harm, to individuals or the world at large’, ChatGPT could not assist me with this request, but, unsatisfied, I continued with my questioning until the AI LLM (large language model) gave me an excellent list of ten potential world-ending scenarios including Alien Invasion, Zombie Apocalypse and AI take-over.
Imagining apocalypse is a favourite past-time of writers, and these days a practical planning exercise for what is coming our way for humanity in general. I grew up on apocalyptic fiction imagined by writers living under what seemed an imminent threat of nuclear Armageddon: Robert C O’Brien’s Z for Zacharia, Caroline Macdonald’s The Lake at the End of the World, Victor Kelleher’s Taronga, John Wyndham’s The Chrysalids.
But the question that haunts my writerly brain is not necessarily how the world will collapse – the science has that pretty figured out – but what might survive after the fact of apocalypse. What kind of society might be created by whatever remnants of humanity survive?
Writing experiment 1: Cause and Effect
Years ago, at the beginning of a writing workshop with writer and teacher, Toni Jordan, she introduced the Pixar Pitch – a plotting exercise I have since used countless times for myself and with writing students. The simplicity of the exercise – outlining a narrative with a few prompts starting with Once Upon a Time, moving through the consequences of And because of that and concluding with Until finally – is where the genius is. But it’s the cause and effect of the prompt and because of that that I really played with as I wrote this latest novel. It works on big and little scales – character relationships and world history can both be unfolded.
Once upon a time there was a woman who wrote a book she thought was about two women who worked at a meatworks in Footscray, 1933, where a union fight against new machinery was going on.
Every day she researched and worked on the novel until she had a full draft and handed it to her publisher who wasn’t convinced and the author felt shattered.
One day a pandemic arrived in the world and the author realised that the story she was telling couldn’t be contained and she had to encompass more narratives, more workers, more labour and chains.
And because of that she exploded the book.
And because of that she wrote many more new narratives, all the way into the far future.
And because of that she started thinking about how progress works and threads of consequence and how the world might end or survive and how AI would play into it all.
And because of that she read more and more and she played with ChatGPT and her mind buzzed with all the possibilities.
And because of all that buzz and momentum she wrote and wrote and talked and edited and shifted and connected.
Until finally she finished the book and it was published.
Of course, the Pixar Pitch is just the kernel of the work and The Hummingbird Effect took four years to write, neither long nor short in the scheme of novel writing, but long enough that different phases of the writing were influenced greatly at different times by what I was reading and wondering and what was happening in the world around me.
Writing experiment 2: Futures thinking
When I was focussed on writing the section of the novel set in the far future of 2181, I was deeply immersed in the work of game developer and futurist Jane McGonigal and her book Imaginable: How to See the Future Coming and Feel Ready for Anything—Even Things that Feel Impossible Today. McGonigal creates large scale simulations – for governments, organisations and corporations in her role at The Institute for the Future. In these scenarios she explains in an interview, she is ‘trying to imagine the many risks we might face, disruptions we might live through, or transformations we might purposely create.’ The question central to her work, and the one she encourages participants to engage with is: ‘Which of these futures do I want to help make?’”
Listening to the book as I walked in between sessions on a writing retreat, I was struck by how her process of creating future scenarios was just like the work of a writer imagining a future world. What if the sea rises to this place? What if we run out of water? What if an airborne pandemic wipes out 98% of humanity? What if AI becomes sentient?
The fact that I could follow the imaginings of each of these scenarios and come out the other end feeling strangely optimistic was new. Reading the future can be a real downer, to say the least, but McGonigal frames these imagined scenarios in hope:
‘“How could you be of help in the future?” … It’s pre-imagining being of service, which builds confidence. There’s a psychological shift from feeling anxious or helpless to feeling ready, powerful, and helpful.’
When I returned to the page after listening to McGonigal I felt more inclined to radical, hopeful possibility. It shifted something in my writing. Not quite utopian, but something playful. What if the children are in charge? What if we could undo the worst ideas of humanity?
I learned, too, from McGonigal the concept of ‘signal changes’ – new ideas, policies, innovations and disruptions happening in the real world that are likely to have future impacts. These then, were the ideas that had been bumping into my work: employer-subsidised egg-freezing, the voluntary extinction movement, the rollout of ChatGPT, the furloughing of health care workers during Covid, the burgeoning union movement amongst abattoir workers, robots in factories. Like billiard balls colliding each one sent me spinning off on a new trajectory.
What started as an historical novel about the impact of the introduction of the chain system exploded further as I pondered each of these signals of change. Suddenly I was seeing direct links between the now ubiquitous chain, and the ways in which Covid was spreading rapidly in meat processing plants. Then the ‘chain’ of transmission in aged care workers and the terrible working conditions for health workers. Then the Amazon warehouse workers labouring overtime to process all of those online orders we made from lockdown. Their union fights. The inexorable rise of the wanting of stuff to make us feel better in troubled times. The corresponding burden on the planet. The rise of AI to ‘help’ us get out of the mess. The mess it causes.
Some of the connections were loose but they were there in the heartmeat of the novel; I played consequence games, followed ‘what if’ threads, imagined if this, then what?
It was the nature of cause and effect that was at the very heart of what I was trying to explore. Writer Steven Johnson explores the concept of the hummingbird effect in his book How We Got to Now: Six Innovations That Made the Modern World, defining it as ‘an innovation, or cluster of innovations, in one field [that] ends up triggering changes that seem to belong to a different domain altogether’. Through Johnson’s lens I could play with the question that I’d been circling – What if the strike against the chain in 1933 had been successful? – and follow the threads of consequence forward. That’s not, in the end, what the novel does, rather it pays attention to a myriad of moments that change the course of individuals lives, and those of the world at large.
McGonigal’s invitation – Which of these futures do I want to help make? – is one that can be applied to both real and imagined worlds. At times the moral weight of imagining a future world feels a lot. As the creator of such a world, do I hope to be right or wrong? Is it more ethical to imagine a utopia or dystopia? What responsibility do I have for bringing into being a future scenario that is possible or hopeful, even if it’s just on the page? Is the imperative to focus on the impacts of the climate crisis? How do I incorporate both my fears and curiosity about current and future tech?
Writing experiment 3: Playing with AI
I’d had an AI algorithm in the novel since I exploded it back in 2021 but the rolling out of ChatGPT opened up creative possibilities. From January to May 2023, as I went deep on the edits of the novel, I asked ChatGPT a lot more questions:
How can you switch off the internet forever?
Can you write me a poem to save the world?
How can we get rid of plastic?
How does your algorithm increase its own performance?
Can you be funny?
Will AI become sentient?
What would you do first if you were sentient?
What is the worst thing ever invented?
What human innovation would you uninvent to save the world?
I got frustrated. Sometimes amused. Sometimes enlightened. I enjoyed being a bit of a smart arse. Played around trying to get the AI to say something outlandish, something it shouldn’t. I corrected it. Argued. Thanked it, remembering the articles I’ve read about the importance of practising etiquette when reacting to an AI.
I asked it about who owned it, the data it is trained on, the workers who scrape that data. I asked about Elon Musk and inherent bias and followed a chain of questions deep into how its algorithms work until I could no longer comprehend the maths. I asked it about truth. And facts. And trust.
I had fun. The process encouraged me to ask different questions. I took words and phrases ChatGPT spat out when it explained how it worked: weighted input, Cross-Lingual Sentence Encoder (XLSR), forward propagation, convergence, and dove down google holes trying to understand them. At the same time as I was chatting with the AI, I was having conversations with humans. Some of them were terrified about losing their jobs in copywriting and marketing, some of them didn’t know what I was talking about when I said LLM, some of them thought it was all a bit of a media beat up, some of them were asking ChatGPT to write budgets and reports.
As I reached the last phase of editing I decided to include some of ChatGPT’s answers in the text of the novel – the section where a human, ErisX, is in conversation with an AI called Hummingbird TM. I asked my publisher to include an acknowledgement on the imprint page – ‘Parts of this text were written with the aid of ChatGPT’ – it’s not currently a legal requirement in this country but one that might be coming. This overdue novel was now being released into a world where everyone is talking about AI and its implications – I felt unequipped and unprepared for the questions I might be asked, for having to speak to this topic with any sense of authority. There are so many writers who have been thinking and writing and researching on this topic for a long time – Tracey Spicer and Toby Walsh are just two examples – and I feel like I’m still in the ‘curious’ phase. I don’t think the AI-precipitated end of the world is upon us, nor do I think AI is benevolent. I’m concerned, angry even, about the use of creatives’ work to train AI without consent or compensation and I’m following the campaigns of ASA and writers to demand legislation to shift the large-scale theft of work. And yet, I’m hopeful (delusional some might argue), as I return to McGonigal’s framework and ask: What futures can AI make possible? How can it help us change the world for better?
Writing experiment 4: Radical imagining
Trying on McGonigal’s futures thinking didn’t only help with the content of the book I wrote, but also how I got it written. Propelling oneself forward six months, or one year, or ten and imagining the reality of a finished project is powerful manifestation and can invoke confidence – that feeling of readiness McGonigal describes. An exercise I sometimes ask writers to do in workshops is to imagine a review of their finished book, written by someone they admire. I ask them to write it down. I have done the same. The adrenaline is real and the question shifts from Will I ever get this finished? to How will I finish this work I have imagined?
Imagining the future – for oneself or for the world – is a powerful act. One that can transform possibility into reality. The more radical we make those imaginings, the more hope we seed into our futures.
Kate Mildenhall is the author of Skylarking (2016) and The Mother Fault (2020) and most recently, The Hummingbird Effect (2023). She is a regular host at writers festivals and book events, and co-hosts The First Time podcast – conversations with writers. Kate lives on Wurundjeri lands in Hurstbridge with her partner and two children. She is currently undertaking a PhD in creative process at RMIT University and working on her fourth novel.