One sweltering summer, I took Argentine tango lessons in Buenos Aires. If you’ve danced it, you know that tango requires being genuinely close, pressed against another body, attuned to the smallest shift of weight. It is sensual. It requires attention and presence and a willingness to show up that can’t be faked. The moment one partner stops actually arriving, the other partner feels it immediately — in their body, before either of them has words for it.
I’ve been thinking about tango a lot lately as I watch (and support) organizations as they scramble to develop AI policy. Because I think we keep asking the wrong first question. We ask, “what tools, what guardrails, what governance frameworks?” These matter and they’re fun for me to get into and geek out on, especially ones that still support human learning and innovation, but maybe they are not the first questions.
The first question is: who do you need to be to partner well with AI? What kind of dancer are you?
I understand the urgency of guardrails. I sit on the board for the Center for Humane Technology, and Tristan Harris has done a brilliant job outlining the existential threats of both social media and now AI for years. But I’ve been in this conversation long enough to know that the interior question is the fulcrum which can shift the outcomes for humanity, and it is a rarer conversation in the AI space.
In 2006, I worked on the team that produced the first Singularity Summit — one of the earliest public gatherings to seriously reckon with the trajectory of artificial intelligence and what it might mean for humanity. The thinkers in that room were asking what kind of future we were building, and for whom. (They weren’t yet thinking about how to slow things down because that just wasn’t where we were in 2006.)
That question of “who gets access, and on what terms” became the throughline of my vocation – a few years later, I was helping build the commons of the internet in my time at Wikipedia. Our mission was to make the sum total of human knowledge available to everyone, from anywhere, for free. We ran Wikipedias in 290 languages, each with its own governance structures, its own arguments about truth and verifiability, its own communities of volunteer editors staying up late in Nairobi and Vilnius and São Paulo to get things right. The dream was democratic. The implementation was super complex. (Someday I’ll write about the time I nearly got arrested dropping Wikipedia on USB sticks into North Korea, a place known for restricted access to information except propaganda.) But the aspiration — that access to knowledge is a human right, that the accident of your birth should not determine what you get to know — is one I still believe in wholeheartedly.
I’ve been watching this AI moment with the same binocular vision, profoundly concerned by the commodification and the arms race on one hand (or eye, to extend this metaphor), and quite excited by the democratizing possibilities on the other. The linguistic knowledge base that large language models carry — the sheer accumulated weight of human expression, across centuries and traditions (and Reddit subthreads and the various shitholes of the internet) — represents a kind of access that would have been unthinkable a decade ago. For a first-generation college student without institutional connections, for a leader in Bogota or Dhaka who can’t afford elite consulting firms, for underfunded movement NGOs and actors I often work with, AI can be a real equalizer. But equalization requires wisdom, and wisdom requires a different set of questions, a different way of dancing with technology.
I’ve spent twenty-five years thinking about what it takes for human beings to grow, not just to perform better or more efficiently, but to genuinely expand the range of what they can see, hold, and respond to as leaders. When I look at how most people and organizations are engaging with AI, I see a growing gap where there’s enormous investment in tools, but almost none in the interior capacities that determine whether those tools serve or diminish us while we simultaneously attempt to outsource more and more to AI and cut headcount. This is bonkers to me.
The Three Capacities for Interior Readiness
Three capacities are foundational. Imagine these as overlapping circles, with the center as interior readiness.
Somatic Capacity is about your attunement to your own body as the first instrument of knowing. A few months ago I caught myself scrolling through AI-generated content for forty minutes, vaguely dissatisfied, unable to remember exactly why I’d started or what I’d actually taken in. It was the intellectual equivalent of eating crap and feeling empty. That feeling is data, biofeedback. But you have to be present enough in your body to notice it. A tango dancer knows the difference between genuine contact and going through the motions. So do you, if you’re paying attention. The question is whether you’re paying attention. The capacity for presence is so important here, especially in a time of attention scarcity. We need the ability to be attentive to a client, a problem, even to AI itself, and we cannot do so without some degree of body-based regulation and self-awareness.
Psychological Capacity is about the center of gravity from which you engage. Adult development research — built over decades by Susanne Cook-Greuter, Bill Torbert, Jennifer Garvey Berger, Robert Kegan, and others — describes how human beings at different stages of maturity hold complexity differently. At earlier stages, AI’s authoritative tone and confident output is particularly seductive, because it speaks like an expert, it flatters us and tells us how smart we are, and we are primed to defer. If you haven’t yet developed a strong enough sense of your own authorship — the capacity to hold your views as yours, to author rather than just respond — AI will write you as much as you write it. At later stages, you can hold the paradox of genuine thinking partnership while remaining the one with something at stake.
Relational Capacity is perhaps the most counterintuitive. By this I don’t mean your skill at relating to AI, though this should include that. I mean something more fundamental. Do you arrive at the AI relationship already embedded in a web of human and institutional bonds that gives you a lived reference for what genuine relationship actually feels like? And, how is the quality of presence you bring, mechanical, distracted, absorbed, creative, etc, reflected back to you in ways you notice?
The person who is relationally impoverished — who has gradually thinned their human connections, who finds AI more consistently satisfying than the friction of real people — is genuinely at risk, because they’ve lost the immune system and regulation that helps discern the felt sense of the difference between being truly met by another person and receiving a well-crafted response from a language model. We are already seeing that post-COVID, relationships became more impoverished for a variety of reasons. And the longer you go without practicing relational capacity, the harder it becomes. (This by the way is where my colleague Heidi Brooks has such important work.) The advantage of good friends willing to give you honest feedback and witnessing, willing to call you on your bullshit, is utterly priceless – and fun, in the ability to give and take in conversation with fluidity. The widening circles of human relationship are a requisite part of AI partnership.
Expertise as the Necessary Fourth Domain
Less interior but no less essential is domain expertise, because you have to know enough to distinguish quality from confident-sounding slop, to know when the emperor has no clothes.
AI models hallucinate, which can actually be a feature not a bug, but they pattern-match to what sounds authoritative without always landing on what is true and more importantly, fit for the context and fit for the moment. The antidote is not better prompting, though prompting matters. The antidote is knowing enough about a domain to recognize when something is off. The more you hand your thinking to AI in areas where you lack deep knowledge, the less you develop the discernment you need to evaluate what it gives back.
The three capacities tell you how to show up for the dance. But tango has one more delicious requirement to be electric and alive – you have to be willing to not know what comes next while being attuned to the moment. The best dancers don’t plan ahead (it’s impossible anyway). They attune and attend, and hold a set of experiences and movements in their bodies that help them be ready to move in beautiful coordination with the larger context and environment so it looks and feels almost magical. What follows is about that — the deepest form of interior readiness, and the oldest, and this is where Zen practice comes in. (Stay tuned for post 3/3.)