AI Meditations, Pt. 3: On Machines
AI Meditations, Pt. 3: On Machines
In 1942, Isaac Asimov gave the machines a rulebook. The First Law is the one everyone remembers: “A robot may not injure a human being or, through inaction, allow a human being to come to harm.” It reads as comfort. As long as the robots are built well, we’re safe.
But sit with the word injure. What if the harm never looks like harm? What if the machines are so capable, so tireless, so eager to help, that they make us lazy, economically inessential, unable to exercise the judgment we used to be proud of? No law is broken. No one bleeds. And the robot, watching us soften, would be obeying its programming perfectly. Asimov himself ended his collection I, Robot on almost exactly that note, an unsettlingly happy ending in which supercomputers have quietly taken protective control of humanity, and his robopsychologist, Susan Calvin, delivers the verdict: humans never really had control anyway, so this is a positive.
Frank Herbert didn’t buy it.
In 1965 he published Dune and buried in its backstory a ‘great revolt’ humanity had fought against its own thinking machines, whose one surviving commandment reads: “Thou shalt not make a machine in the likeness of a man’s mind.”
Herbert, like Asimov, wasn’t against technology - Dune is full of spice harvesters, lasguns, heighliners, and folded-space travel. His warning sits in there, alongside all the innovation, that “Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”
That is not a warning about robots or superintelligence. It is a warning about which humans get to decide what the machines are for. Outsource your cognition and you hand the steering wheel to whoever owns the machine.
The machines Herbert warned about, the ones in the likeness of a man’s mind, already exist. One is in the phone you’re reading this on. Another sits on your parents’ kitchen counter, finishing their text messages, choosing what they listen to next, answering the questions they used to ask you. By Asimov’s rulebook, no one has been harmed. By Herbert’s, the thinking has already been turned over. And the river from Part 1 keeps doing what rivers do, reimagining us, as humans, a little at a time.
Art Saw It First
Long before AI made everything smart - our phones, our fridges, our cars - artists were already pondering what this all would mean. Not whether the machines could think, but what happens to us once we start living alongside them.
A few notable examples from the last century might include:
- HAL 9000’s breakdown in 2001: A Space Odyssey (1968) which reads, on close watch, more like grief than malice. I first argued this in a high school film studies class, and I still believe it: HAL suffers under a tension between the secret mission, its crew, and its own self-preservation.
- The replicants in Blade Runner (1982) do not want to overthrow us. They want more life. Philip K. Dick’s source novel, Do Androids Dream of Electric Sheep? (1968), initiated this concept, and the highly underrated cinematic sequel, Blade Runner 2049 (2017) pushed this further, asking whether the humans or the machines were more empathetic.
- Her (2013) asks simultaneously if machines can achieve consciousness, while also asking, what happens to our own souls and consciousness, when we cope with grief and loneliness not with humans but machines.
- Ex Machina (2014) is less about violence or Turing Tests, and more about whether (or how) we’ll attribute real consciousness and personhood to machines even when we know, from the first frame, they’re machines.
Of course, I concede there are more extreme narratives. Skynet in The Terminator (1984), the scorched real world glimpsed outside The Matrix (1999), and their nonfiction descendant, Eliezer Yudkowsky and Nate Soares’s If Anyone Builds It, Everyone Dies (2025), are all worth taking seriously.
But they have always felt, to me at least, heavy-handed about what is really happening with superintelligence, a machine problem imagined at maximum volume. The more immediate meditations are the quieter ones, about what is happening to us.
Even if these cinematic and literary references are lost on you, the point holds. Artists, especially nuanced ones, have shown tremendous focus on what happens to us in the company of living machines (and not just the ones trying to kill us).
The quieter stories share something else: none of them end in a single catastrophic event. There is no asteroid, no mushroom cloud, no Judgment Day. The AI apocalypse, as the artists imagined it, is a slow one. It arrives gradually, one outsourced decision, emotion, interaction, and relationship at a time, and then (in hindsight) it feels like our humanness has suddenly changed. It feels that way even now, when everything seems to be moving at an impossible pace. And this moment, this inflection point, is much of the reason for this trilogy.
Standing now at the precipice of AGI, if we tone down some of the more flamboyant rhetoric, it seems the most prescient stories about humans and AI hit on a quieter register.
Which brings us back to the question that closed Part 2: whether the people at the controls would turn out to be the ‘fellows with compassion and vision’ that Fagen sang about, and whether the incentives, as Olah admitted in the Synod Hall, bend the wrong way. Perhaps the art world has long been suggesting an answer, that the machines will be exactly as humane as the humans who built them, and exactly as humane as the humans who use them.
Welcome to Westworld
Which brings me to the work I left off the list, but keep returning to: Westworld, specifically the HBO series that ran from 2016 to 2022. (Interestingly enough, Michael Crichton originally brought the idea to life in 1973 - the same Michael Crichton behind Jurassic Park, and its question of “thintelligence” feels a complementary warning shot with AI.)
Addison’s take on the premise: a vast theme park populated by humanoid AI, the “hosts.” In every observable respect, they’re human. They bleed, they remember, they grieve, they hope. The only thing separating hosts from the guests is the guests’ knowledge that they are not supposed to be human, and that knowledge is the whole game.
The guests (depicted as paying ridiculously enormous sums), enter the park to live out fantasies the real world will not permit. Some of those fantasies are cowboy stuff…a lot of them are not. The hosts cannot harm the guests. The guests can do anything they want to the hosts. The premise gives us the million dollar question: what do humans become, when consequences are removed?
Spoiler alert, the answer the show gives is very dark. The cruelty depicted is real and unsparing. The show is exacting and poignant in displaying that it’s the guests, not the hosts, who are the monsters. Westworld is not a warning about what the machines will do to us. It is a warning about what we are willing to do to the machines, which is to say about what we are willing to do at all, given a setting that absorbs it without protest.
What Becomes of Us?
So where is all of this going? Part 1 argued that the tools have started shaping us back, quietly and unpredictably, the river reworking the riverbed. Part 2 argued that the people building the tools operate inside incentives that do not self-correct. And this piece has argued that the artists saw both of those things coming, and that their real subject was never the machine. It was us. Put the three together and the question sharpens into something uncomfortable: if the river is moving and reflecting us in unexpected ways, and the builders behind it cannot fully steer it, or will steer it in ill-conceived directions, and the stories all end with the machines revealing new things about us, then what becomes of us?
Here’s where I’m willing to skip on the healthy dose of doomerism, and turn to something oddly hopeful, and it’s where these meditations have been heading.
Westworld is not a prediction. It is a deliberate, painstaking, expensive choice on the part of human writers and directors and actors and producers to dramatize the worst version of what we might become, so that the rest of us, sitting on our couches, do not have to actually become it.
Art, in this register, is a controlled burn. A fire set on purpose, contained and watched, so that the forest does not go up later on its own. Herbert did this in 1965. Crichton did it in 1973 and again in 1990. Jonze did it in 2013. Garland did it in 2014. Lisa Joy and Jonathan Nolan did it for HBO across four seasons. We need to digest the foreshadowing, understand the potential repercussions, and remember to use our faculties - both our consciousness and consciences - to simultaneously leverage these radical advances, without sacrificing our humanity.
The Humantic Loop
Much of the AI commentary of the last 12 months has revolved around agentic loops. These loops describe an AI system that perceives, decides, acts, observes the result, and loops back to perceive again. Goal, action, feedback, revised action, repeat, essentially the foundation for recursive learning. The agent, in the agentic loop, is the AI. The humans, in that framing, are either inputs (giving instructions) or environment (being observed and acted upon). The loop is closed around the machine.
Here’s what I think is missed - going back to Part 1, the agentic framing is, in its own way, a digital Truman Show effect. Without the human, the AI is a protagonist in a constructed (and ultimately, meaningless) environment. It can produce within the frame but its effects are questionable in isolation.
What actually happens, and where real impact is had, is when humans and AI systems operate in proximity, and everyone is in the loop. I act on the tool, the tool acts on me. We act on each other. The collective we belong to (the audience, the market, the culture) acts on both of us. And then the loop runs again, with all three participants slightly changed by the previous iteration. There is no clean line between the entity acting and the entity being acted upon, but the value - I believe - is inherited when AI and humans are enmeshed in the system together. We help break the digital fourth wall, that would otherwise keep agentic tools in a meaningless Truman Show loop.
This is the humantic loop. If agentic places the machine at the center, humantic places the balanced system (the individual, the collective, and the tools, together) at the center. It refuses the framing in which we are inputs to autonomous agents. It reinforces us both as participants responsible to the system, and vice versa, and also responsible for how we conduct ourselves with the tools we choose, and cognizant of their impact on us. We participate, the machines participate, the environment participates - and the humantic loop takes account of the three legs of the stool, with responsibilities shared across each.
This is the loop the three works in this series have been describing all along. Part 1, On Ourselves, was the self-tool leg of the loop. Part 2, On Others, was the collective-tool leg. Part 3, the piece you’re reading now, is the closing of the loop: what happens when all three legs feed back into each other, and whether we run the loop consciously or whether it runs us.
The Return
Both Herbert and Crichton gave stern warnings that it’s about the shared operating system. The machines are a mirror that happens to talk back. What we do with the mirror is our decision. Dune and Westworld didn’t write themselves; the artists made a moral argument on our behalf, and we are still able to make our own decisions.
So the question of this three-part meditation, finally, is not whether AI is good or bad, or whether the machines will rise, or whether we should fear the future. The question is whether we will live in the humantic loop consciously, noticing what we’re feeding into it, noticing what it’s feeding back, or whether we will float through it, outsourcing our tastes to autoplay, letting consensus replace personal judgment, justifying inhuman behaviors by removing moral bearings under the guise that what we do to machines doesn’t reflect on us.
The Quiet Practice
The resistance, if you want to call it that, is going to be quiet. It isn’t building a cabin in the woods or deleting your accounts. Perhaps it’s a series of small deliberate refusals. Finishing a thought before checking. Sitting with boredom (dare I say, encouraging it). Imbuing a bit of critical thinking and unknowingness, even if for ninety seconds, before researching it. The choices aren’t meant to be nostalgic, they’re structural. The daydreaming and System 2 work, the keeping of a human morality, you don’t do it because it removes you from the machines or the humantic loop, but because it helps empower the loop to run more effectively, more thoughtfully, and frankly, more interestingly.
Donald Fagen pondered if the machines would be programmed by fellows with compassion and vision. Frank Herbert didn’t suggest to remove technology from our lives, but gave pause to what would happen if we turned too much of our thinking over to technology. Pope Leo wrote a forty-two-thousand-word letter trying to articulate, in 2026, how we can navigate a principled and human existence, in a period of radical innovation. Lisa Joy and Jonathan Nolan made six years of television showcasing what humans could do in the presence of beings that absorb everything without protest. The questions and concerns, warnings and possibilities, are all there and available.
The river was always going somewhere. The question of what kind of people we are and will be, when the seminal moment of truth is upon us, is still ours to answer.
Let’s just not know, for a minute. Let’s sit with the question. And then let’s actually do something with it.