The Unoptimizable Things
On what holds when the systems around us stop having limits
At the rate things are changing, the average lifespan of one individual human can start to feel insignificant. A rounding error, almost. Measured against the projected operational window of a large language model — or against cosmological time — eighty, ninety, a hundred years sounds modest. Maybe meager. But what goes underrated is how much you can actually do in that span. Measured against the experiences of a life being actually lived, it’s an enormous canvas.
We’ve absorbed a cultural habit of treating limitation as a core problem. Bounded time, fallible memory, bodies that need sleep and food and rest — in the productivity literature, the longevity “#dontdie” craze, and now throughout the AI capability discourse, all of these show up as design flaws. What we as humans are slower in doing or worse at becomes some kind of cosmic measuring stick that determines our worth.
But — and this part tends to get lost — limitation is also where values become visible. In a moment when so much feels outside our control, that’s worth pausing and reflecting on. Choices are only real because they cost something. A commitment means something precisely because it closes off other paths. A relationship matters, in part, because it is bound to have an ending. Constraint is what turns preference into priority — and priority is how you find out what you actually believe.
This is less obvious when everything feels abundant. When time feels infinite, nothing is urgent. When attention has no cost, everything gets a little of it. The restrictions, the boundary lines are what make any of this legible.
There’s a version of the AI capability argument — the dominant one, at the moment — that demotes the human brain by measuring it incorrectly. Or at least ungraciously. The brain is, granted, not optimized for raw computation — it’s slower than hardware, less consistent than software, prone to fatigue and bias and motivated reasoning. None of that is in dispute.
What the brain is optimized for is more complex and harder to name with precision. It moves a body through a physical world. It integrates sensation — temperature, rhythm, the subtle acoustics of a room, proximity to other bodies — into continuous interpretation. That interpretation becomes meaning: language, attachment, sensory delight, the felt difference between a room full of strangers and a room full of people you love. Speed benchmarks don’t capture any of that processing, because the benchmarks weren’t designed to.
Even if AI eventually outperforms us on every task we currently test for, that doesn’t settle the question of what the brain is for. We built those benchmarks. We decided what to measure. (As I often say: Machines are what we encode of ourselves.) Those decisions reflect our assumptions about intelligence, and our assumptions have always been narrower than the fullness of the phenomenon.
And yet the self doing all this meaning-making is less unified than we tend to assume. The human body is a collective — a walking collaboration of trillions of organisms, of which the cells we’d identify as specifically “human” make up something closer to a tenth of the whole. The rest: bacteria, viruses, fungi, archaea — the microbiome and immune ecosystem running largely outside conscious awareness or direct control.
You can tend this system or disrupt it, feed it well or poorly. But governing it, in any precise sense, isn’t available to you. The gut makes decisions the brain doesn’t direct. The immune system runs its own protocols. The biochemical processes that shape mood, cognition, and physical resilience proceed with considerable autonomy from whatever you think of as the part that’s “you.”
This is a different register of limitation than finite time. A bounded lifespan feels like a ceiling — a constraint on how much. This is more vertiginous: the “self” doing all the accepting is itself a provisional arrangement, a temporary coordination among systems that mostly don’t report to it.
Humans have always lived inside this. The body’s relative autonomy has been a fact for as long as there have been bodies; what’s changed is our capacity to describe it. Learning to hold that without panic — to recognize that the self was always a coordination rather than a command center — turns out to be one of the more durable achievements of a human life.
What it adds to the picture is a scale argument that runs in both directions. We are exceeded by systems larger than us — institutions, history, the accumulated weight of what came before — and inhabited by systems smaller than us that predate us and largely ignore our preferences. Every human being sits in the middle of two kinds of limitation, not at the apex of one.
The same dynamic appears when we build systems at scale. The person whose situation is an edge case, whose experience doesn’t fit the model, whose needs are too particular to register in aggregate data — they fall through the cracks for much the same reason conscious identity can’t govern the microbiome: the governing system isn’t operating at the right resolution to see them. Whether we treat that as an acceptable tradeoff or a design failure is, again, a values statement.
The systems larger than us raise a separate question, because we made them. We’ve always built structures that exceed us — institutions that outlast any individual, projects that continue past the people who started them. The work of being human in a society has always involved entrusting things to systems larger than yourself and hoping the values embedded in those systems are ones you’d recognize.
What’s shifted is the degree, and the degree has become hard to minimize. AI systems now exceed human cognitive reach across domains that were, until recently, understood as definitionally human — judgment, interpretation, creativity, pattern recognition at scale. What we choose to do with that capacity — what we automate, what we preserve for human judgment, what we refuse to optimize because optimization isn’t actually the point — is being expressed whether we’re deliberate about it or not. The choices are already legible — whether we’re making them consciously is the open question.
Limitation, read this way, is one of the conditions for meaning. The finite life is the one where things are actually at stake — where a year spent matters, where what you make with your particular span of time and attention and capacity is yours in a way that has some weight to it.
When the systems around us stop being limited in the ways we are, the question of what we choose to do with that doesn’t get smaller — it gets louder.
And the fact that you’re asking those questions — what should we protect, what deserves human judgment, what should stay unoptimized no matter how inconvenient that is — is itself an orientation and a declaration.
What holds, when the systems around us stop having limits, is what has always held: the weight you assign to things, the choices you make within your particular constraints, the orientation you carry into a world that has never waited for anyone to be ready, that you did not design and cannot fully govern — and never could. That's also, and always has been, an invitation, if you choose to see it that way, and an opening. What you carry into it is yours.
About Kate O’Neill:
Kate O’Neill is widely known as the “Tech Humanist.” She is a global keynote speaker, and award-winning author of What Matters Next, A Future So Bright, and Tech Humanist. As CEO and Chief Tech Humanist of KO Insights, she advises Fortune 500 leaders, governments, and organizations on making technology decisions that serve humanity—not just markets. Her broader work explores the intersection of technology, humanity, and meaning — ensuring that as we build increasingly powerful systems, we keep human dignity and democratic values at the center.
Named to the Thinkers50, the global ranking of top management thinkers, Kate has been featured in WIRED, the New York Times, and the Wall Street Journal. She hosts The Tech Humanist Show, exploring how technology shapes human experience.
Learn more at https://www.kateo.info/.
If this piece resonates with you, share it. The algorithms may not always reward nuance and hope, but humans still do. Let’s make sure the right message spreads.