Escaping the automation
Channelling frustrations from conversations and reports on AI
It is the end of March 2026 and I am weighing up the benefits of an AI assistant. I have ‘to do’ lists in my phone, my google docs, my work notebook, my journal. I have a paper calendar on the kitchen wall where my partner and I keep track of plans. I have three different email-related google calendars, as well as an iCalendar. I have 2 social media apps, plus WhatsApp, plus Miro, and Notion. As someone balancing a weekday job with freelance work, writing and co-leading a local community of futurists, it’s hard to keep tabs on it all.
I have a friend whose partner made and sold a software company. He has since built them both an AI agent and given it a debit card: it keeps tabs on their commitments and chores, but also scans movie listings, gig tickets, restaurant openings, so they might be “out more” and “doing things they want to do”.
An assistant that combines work with home and culture and social activities sounds increasingly interesting.
It’s hard, though, to draw the line between boring admin and vital/necessary work.
Previously, combing through cinema listings was part of the fun of discovery, the consideration of each is part of the process that helped me shape and hone my own personal taste in film. I remember reading The Guardian’s The Guide cover-to-cover when I was a kid at the kitchen table, never mind the fact I lived in the countryside and couldn’t access arts centres myself. It made me feel like things were happening out there. I was excited to grow up and be a part of the big, wide world.
In the evenings, revising for exams before Spotify, I’d listen to late night radio and write down on scraps of paper the names of songs I’d heard and liked, band names like “XX??” that I’d find later in the gig listings. It’s how I found myself at Reading Festival in August 2009 watching three shy people dressed all in black play Intro.
However much I feel tired from constantly shifting tabs between my work, home, social and cultural lives, I also think that, hidden amongst the admin and the information hunting, lie the micro-decisions that teach us and, cumulatively, shape our interactions with the outside world.
How can we possibly draw the line between useful and un-useful friction?
I started this post talking about calendars, I want to end it talking about war.
In February in Iran, a US drone attack made with software running on outdated information killed over a hundred schoolgirls. In the Guardian last week, writing about their deaths, the author Kevin T Baker (who has a substack), placed this error in the context of decades-long history of military automation.
Baker discusses how the bombing happened after the US military sought to use technology to eliminate the (in their eyes) mundane and wasteful task of watching thousands of hours of drone footage in order to identify the “right” targets. Setting aside the question of what target could ever be “right”, Baker makes it clear in his essay that the task of reviewing video also provided the necessary time delays/friction/checks and balances that could have spotted the mistake.
It’s the clearest articulation of concern I’ve yet seen about what processes AI is smoothing, and whether we can ever understand the consequences of what such smoothing might be, before it is too late. We can retroactively identify the “bad” friction of video-watching as life-saving “good” friction of caution, over-eagerly yet mistakenly eliminated. Even so, this 20/20 hindsight perspective is too late to bring back the schoolchildren in Iran.
In his eulogy for manual work, Baker’s essay conjures up an alternative concept of agency. A long chain of checks that, rather than slowing things down, in fact results in an ecosystem where everyone has confidence in their labour, takes pride in their work, develops their skillset. Perhaps, outside of the US military, in a more democratic system, they might even have their recommendations listened to.
In a world that’s increasingly automated (from the military to song playlists), when I encounter pockets of real human agency within the system, I feel strangely elated. Last week, I ran to catch a bus in central London that left just as I arrived. The bus behind me pulled up, the driver beckoned me inside, said “We’ll try to catch it up!” and let me ride inside for free as we raced through Holborn.
I wonder if, in a less process-optimised society, with everyone thinking on their feet, making decisions based on the information in front of their actual eyes, we might move less by instruction and more through shared knowledge and empathy. I wonder if there would be less jobsworth shoulder-shrugging, less blame on the “system” and more space for more serendipitous individual agency-taking. And - I am not sure that delegating our own “agency” to an AI “agent” takes us any closer to bringing this world into being.
*
I write weekly on nature, culture and technology and am always open to chat more about ideas here or projects to collaborate on. Hit “reply” and you should reach me.
Did someone forward this to you? Drop your email below to subscribe for free.