Among the latest digital preoccupations is an obsession with agency, specifically, the value of being high agency. Both employees and bosses should be high agency. The most important thing you can do for your kid is raise them to be high agency. If you aren’t already, you had better start asking yourself if you are high agency.
As far as I can tell, being high agency is tech speak for feeling empowered to act and having the initiative to do so. The high agency individual sees a problem and just solves it. He* stumbles upon an opportunity and seizes it. He sweeps his gaze across the world and sees not endless barriers of laws and norms and material constraints but possibilities for invention and profit.
*Discussions of high agency individuals skew heavily male, though not exclusively so.
If this sounds familiar, it is because the high agency man contains aspects of quintessentially American archetypes. He’s an echo of the pioneer, the entrepreneur, the wildcatter and the iconoclast, which is to say, the high agency man is in part the latest example of America’s abiding love for the confident individualist.
Like individualists of all stripes, the high agency man opposes the constraints of law and custom, which is a good impulse in moderation, because both bureaucracy and habit can impede useful action. The latter may not be so much of a concern these days, but in 2025 the sheer number of rules we all live with can be stifling. Away from the political fringes a consensus is emerging that regulations have become so numerous and excessive that they’ve made doing anything from selling the food you grow to building a small apartment building to laying a mile of high speed rail impossible.
But I think the quest to be high agency also suggests a more timely anxiety, namely, the way most of us have our agency hobbled not by a stultifying government or school system obsessed with conformity, but by the technologies we invite into our lives. Agency may require the boldness to start a new business, but it also requires the self-control to put down your phone. The opposite of agency is passivity, and a great deal of modern technology works hard to cultivate a state of passive consumption in the user. Agency is not just about bucking external forces but about managing internal responses to an overstimulating environment. It rests upon an almost ascetic capacity for resisting all the fun, enraging, interesting distractions that our phones are eager to feed us. The goal is not mere decisiveness, but an almost robotic monomania once the decision has been made.
This brings me to the word agency itself, which has been used for the past couple years not to describe the ideal human, but to describe the ideal artificial intelligence. The AI models currently available may be able to write and research and code far faster than any living human, and they may be able to do so better than most any average person, but they remain fundamentally limited; they need a domain expert watching over them, making sure the writing coheres, that the AI code fits into the larger project, and that the research does not contain fabricated studies or other egregious errors.
An agentic AI would have the fidelity to its goal, or the intelligence, if you prefer, to understand the spirit of the instructions it's been given. It would be able to act with a much broader scope than current models, but it would not descend into delusion or make mistakes. It wouldn’t just research flights for you. Instead, you could trust a properly agentic AI to book you the best ticket to Nashville and trust that it would opt for neither the 26 hour flight with three layovers because of its low price nor the $3500 first class tickets because they’re the most comfortable and direct. Instead, it would select the precise pragmatic compromise between cost and comfort that you yourself would make.
That’s a mild example, so here’s a slightly more dramatic one. An AI could read the day’s news, literally all of it, then propose ten potentially interesting pieces of analysis. It could ask you which you want to read, then provide those pieces with a slant and style calibrated for your enjoyment. In other words, this sort of agentic AI would replace a human in a more direct way than by improving efficiency. Instead of letting one writer do the job of two at whatever daily magazine or blog you prefer, it would do the job itself, with no human other than the end consumer in the loop.
What I find strange is that the people obsessed with agency are maniacally building the technologies that most threaten it. The people advocating for high agency are building both the distracting tools of social media and slop that make it hard to focus and the more ambitious sorts of AI that have an explicit goal of replacing human labor and human creativity. In the near term the threat arrives on our screens as better forms of what we already have — more enticing social media, digital friends who are always and forever in our corner, tools that will let you offload the work of thinking and writing and make you feel like a genius for doing so.
But the long term goal is to dissolve the material world into the digital to the greatest extent possible, and then to create programs that manipulate that digital world with a facility no human will ever be capable of. Whether five or thirty years from now, it’s hard to see how people, no matter how high agency, will have a prayer when matched against agentic AI, at least when it comes to anything digital, and perhaps when it comes to quite a lot of other stuff.
If this makes it sound like I think the headlong rush to cultivate high agency as the trait of the future is actually a race to see who can get stuck circling the cul de sac first, I suppose I do. But that’s mostly because this conception of agency invariably describes a capacity to economically compete in the precise areas AI looks most likely to rule.
I’m actually quite fond of the idea of agency in a more capacious sense of the word, and I think most of us would do well to cultivate it. But rather than focusing on high agency I would embrace human agency.
Perhaps it’s wishful thinking on my part, but lately I have had the sense that questions of how to live well in our technological age have been moving from the abstract to the concrete for a small but growing number of people. This process usually begins in the form of a negative — cell phone bans, tech fasts, pledges to abstain from using AI for this task or that. Speaking as an enthusiastic supporter of phone bans who has severely curtailed my own phone use and has vowed to never use AI to write, I think laying down rules, both personal and collective has to be part of the process. But every bit as important is to not just articulate but to live in a meaningfully different manner from the digital default. Human agency consists of embracing uniquely human existence.
The problem with this framing, with irreducibly human patterns of life set against algorithmic existence, is that the fight looks so unfair. On the one side we have compounding technological advances, the wealth of multinational corporations, the genius of a million engineers, and governments determined to never be a single step behind their rivals. On the other side we have what, exactly? Going on a walk with a friend on an autumn afternoon? Cooking a really good dinner? Kicking around a soccer ball? Singing?
Yes, that is what we have. Human agency comes with stringent limits. Unlike the pure, focused efficacy dreamt of in Silicon Valley, it is in large part defined by its limits. At its best, human agency consists of trying to live well with each other in the world as we find it, to change what we can for the better and perhaps learn from what we cannot. Giving one of my kids a hug is a small thing, but it’s something a computer will never be able to do.