The AI Discourse Is Broken

1. The Three Camps

Something is wrong with the conversation about AI, and it isn’t the technology.

The models work, and they keep getting better. What's coming in the next few years is going to be stranger than most people in this space will say out loud. And I don't think that's because they're hiding anything. The honest answer is just hard to sit with.

But the frameworks people are reaching for to make sense of all this — those are failing us. They're failing us because none of them have a working account of what a human being actually is. That's the missing piece. Not intelligence or sophistication, but a shared sense of humanity. You cannot navigate a transformation of this scale without one. You’ll be lost before you start, and no matter how confident you sound, eventually everyone who follows you will be lost too.

Three camps dominate the public conversation, and each fails for a different reason.

The first is the doomer camp: AI is an existential threat, the technology should be slowed or stopped, and the people accelerating it are either naive or sociopathic. There is something real in this concern (the risks are not made up) but the conclusion is a kind of despair masquerading as responsibility. What the doomers are really saying, beneath the technical arguments, is that they do not trust human beings with powerful tools — that our species has forfeited the right to set its own future.

The second camp is accelerationist, in its various flavors — techno-feudal, hyper-libertarian, e/acc, whatever name it is wearing this month. Here, power is the organizing principle. Move fast. Don’t balk. The first to AGI wins not only the race – but control over the future. A lot of very smart people have talked themselves into something that, when you actually look at it, is really just feudalism with GPUs. Lords and serfs, but the lords live on private islands and well-armed 500-acre compounds.

And then there is a third group, smaller yet harder to correct because they mean well: the complacent optimists. These are the people who will tell you that technological change is basically good, that AI will make our lives easier and full of leisure, that we should be excited rather than afraid — without being able to tell you why, exactly, or on what foundation that confidence rests. Vibes-based humanism, if you will. But when things get hard (and yes, they will get hard) you cannot hold the line with vibes.

If these are the options, no wonder many of us don’t know what to think about AI – if we allow ourselves to think about it all.

---

What I have come to believe, through a path that was not exactly straight, is that this is not primarily a crisis of technology. Rather, it is a crisis of reason. Perhaps even of metaphysics. People are reaching for AI frameworks without having settled the prior question of how human beings should relate.

I’ve been navigating questions of how to relate to people from wildly different worlds from the beginning of my career. I grew up in a DC public school where the joke (a true joke) about your options after graduation was either “Yale or jail.” I made it to Duke, then MIT, but always kept an outsider’s perspective. To me, technology was a way to bridge gaps between worlds. I took what I learned about tech to places that needed it–to Rwanda and Cambodia. I learned what I could from the corporate track at Red Bull. And finally I struck out on my own, building startups on mental health and medical devices. 

In short, I built things and broke things and eventually ended up — by a sequence of events that felt at the time like randomness and feels in retrospect like something else — standing at the intersection of human-centered philosophy and frontier AI.

There was a moment that crystallized it. A conversation with a priest — I’ll come back to this story another time, because it deserves more than a paragraph — but the short version is that something in that dialogue just clicked. In a moment I felt the possibility that reality is not arbitrary, that it has a structure oriented toward something, and that our job — as human beings, as entrepreneurs, as people building things with these remarkable and terrifying machines — is to get better at perceiving that structure of reality and moving with it rather than against it.

So let’s start with what each camp gets right. The techno-feudalists are right that power is a real force. The doomers are right that the stakes are intense. And the optimists are right that a broadly beneficial future is possible. But power with no purpose is appetite, fear without any ground to stand on is paralysis, and optimism without responsibility becomes passivity. The question none of them is asking is the one that matters, perhaps the only one: what is all of this for?

I have an answer…Or the beginning of one. And it begins with an old word that has been either forgotten or so encrusted with age that we might not really know it.

Caritas.

In a simplistic sense, it means “Love.” The love of other people, or charity. More precisely, it signifies the ordered movement of the will toward what is really, truly good for another human person.

This necessitates a willingness to think about what someone really needs, independent of your own agenda, before you decide what to offer them. That requires careful attention and an authentic humility.

It sounds soft. It’s actually the hardest thing there is. And it turns out, I believe, to be the only durable organizing principle for an AI economy.

2. On Love and Knowledge

I'm not saying caritas makes people nicer, that nicer people build better companies and so we should all be kinder. That argument is true but it's also useless. It doesn't ask anything of you, and it leaves you with nothing to hold onto when the hard calls come.

What I am saying is this: caritas is the only cognitive stance that lets you perceive the full frontier of possibilities. And in an economy that is most vibrant at the frontiers — of knowledge, of capability, of possibility — this is actually a competitive claim.

The philosopher Jacques Maritain has an important concept that pertains here: connatural knowledge, or intuitive understanding that comes from life experience. It’s the wisdom that comes from failure–that great instructor–and from the long practice of caring about the world beyond yourself.

A great wine grower knows the harvest by the hands, the nose, the palette, and the memory of dozens of seasons. A great diagnostician knows the patient before the tests come back. All of us have domains where we have gathered connatural knowledge.

What this gives you, in the context of AI, is the capacity to recognize what matters and what doesn’t to yourself, and to other humans. Which paths are worth anything, and which ones aren't. 

Because it arises from life, this capacity cannot be handed over to a lifeless machine. Such knowledge cannot be specified abstractly, in advance, which means it cannot be encoded. The human with strong connatural judgment has  to stay in the loop because they are the only ones who have any idea what the loop is for .

As the tools get better (and they are getting much better and also sharper and more dangerous, as tools do) the task in almost every knowledge-work domain shifts from execution to direction. The person who can seed an AI agent with the right orientation toward a problem, who can recognize a promising path when it emerges, and who also can decide what NOT to pursue — that person becomes far more valuable.

This is especially important because technology, as Hong Kong philosopher Yuk Hui writes, is never neutral. It is always already shaped by the philosophical ideas and cultural assumptions of the civilization (or subculture) that wields it.

You can’t have a technology without a cosmology, even if that cosmology is never stated. Silicon Valley, though not a civilization, certainly has a cosmology. From having worked there, and knowing some of the players – I would observe that this worldview is mostly unconscious. When those assumptions go unexamined, they get baked into the code, and then scaled.

The alternative is not to slow down. It is to get conscious about what we are actually computing toward.

---

Caritas is the inclination to make this work not only for yourself. Genuine attention to what someone actually opens up problem spaces that self-interested people cannot see. The frontier is mostly made of other people's unmet needs. The most rational long-run strategy for building anything of value is caritas—love, if we want to stop dressing it up.

There is a fraternal correction I want to offer here to the complacent optimists — the well-meaning technologists who will tell you that everything is going to be fine; we’ll all soon have flying cars and voice-activated butlers and a social safety net that ensures nobody feels a bump.

Yes, I believe there is a version of the AI future that brings benefit to everyone. But there is also a version of this optimism that is misguided. It tells people that nothing fundamental is changing, that they can absorb these tools into their existing ways of living without disruption, that the transition will be seamless. This is a form of gaslighting – the most well-intentioned gaslighting, for sure, but gaslighting still.

Transformation, for all its use as a buzzword, isn’t easy. It requires entering a kind of pressure cooker: a period where the old ways are failing and the new ones are not yet fluent, and you have to stay in that discomfort long enough for something real to change. What comes out the other side isn't merely an upgrade, but a different thing entirely.

The word for getting ahead of that disruption is caritas. Attention, and genuine interest in the neighbor. The willingness to care about what is actually needed in addition to what’s going to be profitable.

And from that orientation, something – perhaps a society, a community, a communitas -- can be built.

3. Build Your Guild

I want to speak first to two people I have in mind as I write this, because the abstract case is not actually what either of them needs to hear.

The first is someone in the middle of a career — ten, fifteen years in, skilled and serious, with a growing suspicion that the track they are on is not going to take them where they thought it would. They have watched the AI conversation from a slight distance, drawn to the energy of it and repelled by the nihilism of its loudest founders. They have a bad feeling in their gut that they cannot quite name. Trust that feeling. It's your own connatural knowledge telling you the path you're on wasn't built for what's coming.

The second person is earlier in their career — new enough that the tracks they were on are vanishing before they have had a chance to walk them, frustrated and occasionally despairing. What I want to say to this person is that the evaporation of the old tracks is not the problem it looks like.

But you do not have to do it alone. That is the entire point of the guild.

---

I want to tell you about a longtime HR leader who saw the need to change. She had spent twenty-five years in HR, the last ten at an industrial services company — competent, steady, and increasingly aware that the AI tools arriving around her were going to transform what competence even meant. She understood perfectly well what was possible. She just couldn’t find the bridge between what she knew and the AI-fluent expert she was supposed to become.

When my company (my guild) came in to help, we didn't give her a training program. We found the one process in her organization where applying these technologies was so obvious and so bounded that success was nearly guaranteed. I walked her through the starting moves and gave her a pointer for running the rest herself. By the time we reconvened for the formal kickoff, she had already blown through the success criteria on her own. She said something I have heard many times since: "I had no idea that I could do this."

That right there is the whole game. The AI tools are there for anyone to use. What’s truly scarce is the person who has fifteen years of hard-earned judgment, and who’s willing to believe they can actually learn to drive one of these strange, wonderful things.

Most people in her position either don’t feel ready, or don’t know where to start. Getting them to a point where they’re willing to try is half the battle. That’s where the guild can help.

---

The dividing line between labor and capital is collapsing. When coin-operated computers can do the hard capital work, and when human beings with strong judgment can guide them with increasing sophistication, the classic divide between the capitalist – he who owns the means of production — and the prole — the poor sap who operates the machines—breaks down.

Today, labor can act as capital. And capital envies labor’s growing flexibility and skill.

In such a world, the old corporation won't suffice. It was built for a different scaling problem and carries assumptions about hierarchy we no longer need. The gig economy won't do either; it stripped out whatever mutual loyalty was left between companies and their employees.

The form that fits is once again medieval: the guild. A response to feudal times, it bound workers in a trade together in a community of shared craft and skill; in time it organized expertise, reputation, and apprenticeship in a way that aided both craftsman and pupil.

The atomic unit of this structure is the apprenticeship relationship — the bond between someone who has connatural knowledge worth transmitting and someone who is trying to develop it. The guild is what emerges when many of those relationships are in enough proximity to create something larger: collective standards, protocols, the capacity to take on problems that no individual practitioner could handle alone.

Here’s where I must mention that what I am building — and what I am inviting others to build — is a guild of guilds. Our approach is to cultivate small teams organized around specific client transformations, compensated on the basis of outcomes rather than hours, with ownership structures that make workers stakeholders rather than contractors. But our real goal is this: to grow a guild of guilds that empowers individual people to take part – to thrive – in an AI-infused economy.

---

Let’s return to where we started. We are in a crisis of reason, a crisis of confidence that we can make sense of reality and lead it in a good direction. The doomers have resolved this by concluding that we can’t. The accelerationists have settled it by saying making sense doesn’t matter, only speed. The complacent optimists have declined to resolve it at all, which is its own kind of answer.

I keep wanting to call this a third way, but that makes it sound like a compromise and it isn't. It's closer to a refusal… and an invitation. I don't accept that the choice is between shutting the AI train down or letting a handful of people slam it into whatever they feel like slamming it into. There's a third option, which is that we actually pay attention to what's in front of us and to each other, and build a better AI future – our future – from there. I know how that sounds. I'm saying it anyway.

If this speaks to you… good. Don’t stop here.

Take the tools that are available to you now, find one person whose problem you can take genuine interest in, and see what becomes possible when you do.

View AI not as a threat to you, but a chance to build a more intentional future.

And if you want company on the journey, we are here.

The path is steep.

I'm told the view is worth it.