This by no means a definitive list, so don’t @ me! AI is an inescapable subject. There’s obviously an incredible headwind behind the computing progress of the last handful of years — not to mention the usual avarice — but there has also been nearly a century of thought put toward artificial intelligence. If you want to have a more robust understanding of what is at work beneath, say, the OpenAI chat box, pick any one of these texts. Each one would be worth a read — even a skim (this is by no means light reading). At the very least, familiarizing yourself with the intellectual path leading to now will help you navigate the funhouse of overblown marketing bullshit filling the internet right now, especially as it pertains to AGI. Read what the heavyweights had to say about it and you’ll see how many semantic games are being played while also moving the goalposts. Steps to an Ecology of Mind (1972) — Gregory Bateson. Through imagined dialogues with his daughter, Bateson explores how minds emerge from systems of information and communication, providing crucial insights for understanding artificial intelligence. The Sciences of the Artificial (1969) — Herbert Simon examines how artificial systems, including AI, differ from natural ones and introduces key concepts about bounded rationality. The Emperor’s New Mind (1989) — Roger Penrose. While arguing against strong AI, provides valuable insights into consciousness and computation that remain relevant to current AI discussions. Gödel, Escher, Bach: An Eternal Golden Braid (1979) — Douglas Hofstadter weaves together mathematics, art, and music to explore consciousness, self-reference, and emergent intelligence. Though not explicitly about AI, it provides fundamental insights into how complex cognition might emerge from simple rules and patterns. Perceptrons (1969) — Marvin Minsky & Seymour Papert. This controversial critique of neural networks temporarily halted research in the field but ultimately helped establish its theoretical foundations. Minsky and Papert’s mathematical analysis revealed both the limitations and potential of early neural networks. The Society of Mind (1986) — Marvin Minsky proposes that intelligence emerges from the interaction of simple agents working together, rather than from a single unified system. This theoretical framework remains relevant to understanding both human cognition and artificial intelligence. Computers and Thought (1963) — Edward Feigenbaum & Julian Feldman (editors) This is the first collection of articles about artificial intelligence, featuring contributions from pioneers like Herbert Simon and Allen Newell. It captures the foundational ideas and optimism of early AI research. Artificial Intelligence: A Modern Approach (1995) — Stuart Russell & Peter Norvig. This comprehensive textbook defined how AI would be taught for decades. It presents AI as rational agent design rather than human intelligence simulation, a framework that still influences the field. Computing Machinery and Intelligence (1950) — Alan Turing’s paper introduces the Turing Test and addresses fundamental questions about machine intelligence that we’re still grappling with today. It’s remarkable how many current AI debates were anticipated in this work. Cybernetics: Or Control and Communication in the Animal and the Machine (1948) — Norbert Wiener established the theoretical groundwork for understanding control systems in both machines and living things. His insights about feedback loops and communication remain crucial to understanding AI systems.

My thoughts about the Zettelkasten (Slip box) note taking methodology invented by the German sociologist Niklas Luhmann.

Chinese interior studio Various Associates has completed an irregular pyramid-shaped flagship store for drone brand DJI in Shenzhen, China. Located...

AI promises to automate both work and leisure. What will we do then? In 2005, I lived high up on a hill in Penang, from where I could literally watch the tech industry reshape the island and the nearby mainland. The common wisdom then was that automation would soon empty the factories across the country. Today, those same factories not only buzz with human activity — they’ve expanded dramatically, with manufacturing output up 130% and still employing 16% of Malaysia’s workforce. The work has shifted, evolved, adapted. We’re remarkably good at finding new things to do. I think about this often as I navigate my own relationship with AI tools. Last week, I asked an AI to generate some initial concepts for a client project — work that would have once filled pages of my sketchbook. As I watched the results populate my screen, my daughter asked what I was doing. “Letting the computer do some drawing for me,” I said. She considered this for a moment, then asked, “But you can draw already. If the computer does it for you, what will you do?” It’s the question of our age, isn’t it? As AI promises to take over not just routine tasks but creative ones — writing, design, music composition — we’re facing a prolonged period of anxiety. Not just about losing our jobs, but about losing our purpose. The industrial revolution promised to free us from physical labor and the digital revolution promised to free us from mental drudgery. Yet somehow we’ve ended up more stretched, more scheduled, more occupied than ever. Both were very real technological transitional periods; both had significant, measurable impacts on the economies of their time; neither ushered in a golden age of leisure. History shows that we — in the broadest sense — adapt. But here’s something to consider: adaptation takes time. At the height of the pre-industrial textile industry, 20% of all women and children in England were employed, hand-spinning textile fibers. This was in the late 18th century. Over the course of the following forty years, a process of mechanization took place that almost completely obviated the need for that particular workforce. But children working as hand-spinners at the pre-industrial height would have been well past middle-age by the time child-employment was no longer common. The transitional period would have lasted nearly the entirety of their working lives. Similarly, the decline of manufacturing in the United States elapsed over a period of nearly fifty years, from its peak in the mid-1960s to 2019, when a net loss of 3.5 million jobs was measured. Again, this transition was career-length — generational. In both transitions, new forms of work became available that would have been unforeseen prior to change being underway. We are only a handful of years into what we may someday know as the AI Revolution. It seems to be moving at a faster pace than either of its historical antecedents. Perhaps it truly is. Nevertheless, historical adaptation suggests that we look forward to the new kinds of work this transition will make a way for us to do. I wonder what they may be. AI, after all, isn’t just a faster way to accomplish specific tasks; investment in it suggests an expectation for much grander than that, on the order of anything that can be reduced to pattern recognition and reproduction. As it turns out, that’s most of what we do. So what’s left? What remains uniquely human when machines can answer our questions, organize and optimize our world, entertain us, and create our art? The answer might lie in the spaces between productivity — in the meaningful inefficiencies that machines are designed to eliminate. AI might be able to prove this someday, but anecdotally, it’s in the various moments of friction and delay throughout my day that I do my most active and creative thinking. While waiting for the water to heat up. Walking my dog. Brewing coffee. Standing in line. Maybe we’re approaching a grand reversal: after centuries of humans doing machine-like work, perhaps it’s time for humans to become more distinctly human. To focus not on what’s efficient or productive, but on what’s meaningful precisely because it can’t be automated: connection, contemplation, play. But this requires a radical shift in how we think about time and purpose. For generations, we’ve defined ourselves by our work, measured our days by our output. As AI threatens to take both our labor and our creative outlets, we will need to learn — or remember — how to exist without constant production and how to separate our basic human needs from economies of scale. The factories of Malaysia taught me something important: automation doesn’t move in a straight line. Human ingenuity finds new problems to solve, new work to do, new ways to be useful. But as AI promises to automate not just our labor but our leisure, we might finally be forced to confront the question my daughter so innocently posed: what will we do instead? This will not be easy. The answer, I hope, lies not just in finding new forms of work to replace the old, but in discovering what it means to be meaningfully unoccupied. The real challenge of the AI age might not be technological at all, but existential: learning to value the empty hours not for what we can fill them with, but for what they are. I believe in the intrinsic value of human life; one’s worth is no greater after years of labor and the accumulation of wealth and status than it was at its beginning. Life cannot be earned, just lived. This is a hard lesson. Wouldn’t it be strange if the most able teacher was not human but machine?

Weekly curated resources for designers — thinkers and makers.