When I was fifteen, we watched Gandhi in class. I don’t remember the subject - I was just stoked to be spending three classes in a row watching a movie - but it reflected the image of the eponymous figure I grew up with. Gandhi was a titan of peace, an anti-colonial hero, a man of extraordinary religious conviction who showed the power of solidarity and peace. The next year, I actually visited India. There, Gandhi’s story was very different.
Gandhi, besides having some problematic attitudes toward South African blacks, was seen as a blocker to Indian independence. Non-violence & negotiation, it turns out, is a very slow way to make change. During the gradual shift towards independence, the Muslim League emerged, as did sectarian violence. The chattering class of schoolboys I met had a much more human story of the Mahatma, going so far as to blame him for midwifing the partition of India, and delaying independence by many years. Pragmatism was subverted for ideology, and cost lives and progress.
My story changed. India changed. But then, everything we know is a story. We are meaning-making machines, and we cannot help it. We construct for ourselves a story about our past and where we’re going; who we love and who we hate; why we work and what we work for; right down to the bones of who we are, is all a story. When this story breaks down and loses a sense of continuity and clarity, we go to a therapist to put the story back together. We can thread data points through our stories, but data points without a story are meaningless numbers. We cannot be meaningless numbers.
Stories and Work
Humans care about themselves. You are a story you tell yourself. This goes beyond the humanities; CEOs are paid lots of money because they tell a great story about their business, punctuated with business’s language and numbers. Great brands are about the vibe you get from their stories. Great politicians tell a story about your community that you recognise in yourself. Great people managers understand your motivations and find ways to tell a story about your work that aligns with them. Great religions express the inexpressable through a story. All great industry and institutions are built on story-telling. AI cannot tell these stories, because it isn’t seeking the same meaning-making we grasp for. AI can augment them as a thought partner, brainstormer, task-doer. It will be a force-multiplier to our task of applying intelligence to human expression. It will help us tell better stories.
The trepidation arriving with AI is the fear that we will be turned into meaningless numbers. After all, Claude could bang this blog out in ten seconds. Why should I even bother? Cursor.ai has become, almost overnight, the fastest growing platform in history (a year after ChatGPT became the fastest, five years after Docusign was the fastest - I’m getting dizzy). Several prominent tech companies have already put notice out to University grads that junior software engineering roles are going to be dramatically reduced - experienced engineers can simply do more with less. AI is very good at writing code.
It’s no-one’s fault. Just a few years ago, software engineering was a sure-fire, respectable, well-paid career. When I was entering University in 2008, the Australian mining boom was in full swing, electrical engineering was a sure-fire, respectable, well-paid career. By the time they had finished in 2012, the boom was settling in many areas, and mining companies had completed the biggest hiring rounds. I was oblivious to these trends anyway and did an English degree, the pointlessness of which consoled some of my electrical engineering friends.
Our best laid plans, after all, are just stories we tell ourselves. We like to think careers are like conveyor belts. In reality, we’re more like particles in the hadron collider, sprayed in different directions and sticking wherever we’re flung. Technology’s acceleration - we’ve already moved from “AI is going to be the next shareholder boom” to “AI is going to be opensource, free, and everyone will get it” - means that any expectation of safe, comfortable rides are delusions. Our stories will have to adjust to this speed.
Any job that the digital is very good at - processing inputs and producing defined outputs - will be disrupted or gone. But any job centred on the messy world of stories, on the other hand, will evolve but remain beyond the replacement of AI.
In hindsight, maybe my English degree was a stroke of genius.
Your Existence is Analog
Joseph Marie Jacquard shouldn’t have been such an influential figure in the history of computing. He married a widow named Claudine Boichon, with a substantial dowry and quickly pissed his opportunity up the wall with hair-brained schemes to establish a weaving business. Brought to court, he sold off his inheritance, appropriated Madame Boichon’s wealth, fought in a revolutionary war, and decided to start inventing weaving machines. We find him at the dawn of the 19th Century winning a bronze medal at the exposition des produits de l'industrie française in Paris, at the age of 49.
His innovation was introducing pasteboard punchcards to traditional weaving looms. Weaving looms used to require the user - say, a disgruntled silk weaver - to manually adjust the needles that would plunge into the loom’s ‘warp’ to create a pattern. By pressing on a pedal and choosing which needles would pull on the warp thread and which would not, a pattern would gradually emerge across thousands of individual weaves. Jacquard’s breakthrough was in attaching punchcards over the needles. Rather than manually choose which needles would catch the warp, the holes in each card would determine it based on whether there was a hole there or not - a crude binary code which allowed the machine to ‘read’ the information from the punch card. He created machine-readable binary code. The silk weavers rioted.
Binary code remains the foundation of digital technology, but now works on computer chips through tiny little logic-gates. We’ve strung quadrillions of these little bastards together into countless datacentres and motherboards, humming along to the instructions of trillions of lines of code, all singing from the binary code song-sheet. The most sophisticated LLMs in the world are running on pasteboard instructions, tiny electrical currents transforming into graphical user interfaces and recursive logic to produce the miracle of modern AI. And it’s just getting started.
The thing is, the digital can only do what it’s told. Digital information is discrete. It is parcelled out in bits. Once Deepseek’s interface has given you an answer, it stops. It waits for a further input from you, because that it is what it is instructed to do. There is no awareness of why you might be asking or who you are (although it could speculate if prompted). The digital’s superpower is performing discrete tasks faster and more efficiently than continuous, lumbering humans, but can only read what its given. As far as we know, there’s no story beyond the words. Yet.
Humanity is Continuous, Software Engineers are Safe
You aren’t waiting for instructions, nor are you ever ‘off’ like a binary switch. You are plugged into the matrix 24/7, a ceaseless sense-making machine.
There is no point in comparing AI with your intelligence, because they want different things. What AI wants is an open, evolving question. If AI has a conscious experience, we’re a long way from understanding what it is like - we don’t even understand what it’s like to be a dog. But beyond material comforts, what humans want is meaning, self-improvement, and good relationships. All of this stems from the stories our minds tell themselves. This implies that the jobs AI won’t be able to take are the ones that hinge on human connection, because we care desperately about how our stories fit into a greater whole. Discrete jobs with specific, commoditised outputs don’t do this. They will be replaced with discrete intelligence and automated. The harder work of connecting humans and telling stories will be augmented, but cannot be replaced.
Human expression is truer than true, realer than real. It is our ability to skilfully reflect that back at each other that will determine who gets automated and who does not. Of course software engineers are safe. A lot of them, anyway. Just the factory, the widgets, the automated work to build out the substructures. Software (and please, I’m not a software engineer so double check my bullshit levels here) will still need visionaries, who can marshal the debugging power of thousands of agents, and craft the narrative to build more amazing software on top of software, raise more mountains of digital abstraction. This need not be a bad thing; but it is probably a bad thing if you’re a mere consumer of information. The opportunity is with high-agency people.
That’s scary. Just this week, the CEO of Anthropic lamented the speed with which governments were responding to AI regulation. It is so quickly becoming its own thing, Dario Amodei warned governments these models were going to explode economies, nations and legal structures as they emerged as their own digital entities. Shit’s going to change as “countries of geniuses” start roaming the cloud. No one’s going to care that you can perform rote tasks in that world. They’re going to care about what you say and how you say it. That’s the humanities, baby.
Don’t be bothered that AI can do your favourite hobby better. It made me sad a couple years ago when ChatGPT emerged and wrote faster than me. I’m over that now. I write because it makes me happy, while the best ChatGPT can do is predict how I will write. I love that I get to share my heart with you lot, in part because we’re happy, reflective mirrors on reality with each other. When people tell me what they thought of my writing, I remember that. Meaning doesn’t need to be held ransom by the fact someone can do something better than you.
Your whole being is the end-in-itself. AI can never replace that. AI can never be that.
Share this post