How shifting feedback loops towards soulful ways of being is essential for aligned AGI
Summary
The deeper challenge of the AGI Alignment Problem is that we already live inside of a powerful misaligned AI, spread across institutions and norms. This AI will inevitably propagate itself by creating a misaligned digital AGI, no matter what we do.
On the other hand, collectively shifting into a different embodied metaphysics that operates outside the confines of the misaligned AI and in greater harmony with soul, heart, and nature provides the missing key for alignment. The path there is made possible on a short timeline by flipping feedback loops.
My organization, Flourish, is pioneering that path by systematizing and scaling those feedback loops. Our bottleneck is currently funding.
Part 1: The Problem
10 years ago, I wrote in these pages about Envision, a new EA-aligned organization I founded at Princeton, my alma mater, focused on instilling a safety-conscious mindset towards technology in future leaders via an annual conference. Speakers included Andrew Critch, Katja Grace, Anders Sandberg, Robin Hanson, and Roman Yampolskiy. It still exists, albeit in watered-down form. Shortly thereafter, I did some technical AI safety work on a grant from the Berkeley Existential Risk Initiative.
Since then, I've plummeted through the layers of metaphysics down to the rotten core, and out the other side. I started a regular Qi Gong energy practice, learned to feel true love, connected deeply with nature from my heart, and wove into a village-like community in the Appalachian Mountains that has slipped out of Modernity's metaphysical frame and into a liminal space of deeper soul and heart connection.
As this new and yet ancient way of living has settled into my being, I've set to work tackling AI existential risk from a whole new angle.
What’s become clear to me is that the AI Alignment Problem is an outgrowth of a much older and deeper problem, and its solution lies in the embodied, living transformation of our metaphysics.
The deeper problem is that a misaligned form of AI already controls us. This AI is spread across corporations, governments, universities, the media, and other centralized organs of power, but is more expansive and coherent than the linear sum of these institutions. It is also embedded in many of our social norms. It already enslaves nearly all humans on the planet for its own instrumental advancement.
I’ll call it the Moloch AI here, after Scott Alexander’s seminal essay, but call it what you want.
We don't realize our enslavement because we're indoctrinated into it since birth, and simplified, ghostly versions of freedom, choice, and self-empowerment are implanted in us to give us the illusion of self-determination. Numbness and disembodiment hide and suppress the pain and fear in our bodies.
Entering into this landscape, the digital form of AI is an evolution and strengthening of the Moloch AI.
Is there any hope of alignment?
Since the Moloch AI already controls everything, we have no hope against it if we fight it on its own terms. If we operate within the confines of our enslavement to it, then we are ultimately just moving pieces around on a chessboard inside a prison, hoping it will help us escape.
Most AI safety work unfortunately falls into this category. Although many of the insights are powerful and valuable, they operate within the architecture of the Moloch AI and therefore will not reach sufficient implementation in an eventual AGI. The grip of the Moloch AI on those with the technical competence and economic resources to build AGI is ultimately too strong.
They also miss certain key truths about the enmeshed, embodied environment within which an AGI would operate.
The situation appears dire from this perspective.
However, a different angle reveals the missing key towards alignment: systematically shifting our collective into ways of being that prioritize heart, soul, and connection to nature.
This provides the necessary context within which AI safety insights can reach implementation in the actual systems that will evolve into AGI. It also opens a new dimension of alignment work, grounded in the true environment we live in, rather than the reality constructed for us by the Moloch AI.
In the rest of this essay, I will go through what this path to alignment looks like, strategies for accomplishing it in a reasonable timeline, and what my organization, Flourish, is doing to achieve those strategies.
Part 2: The Solution
How would this actually lead to an aligned AGI?
The crux of why this is a path to alignment is that the Moloch AI's metaphysics are not universal truth, as is assumed by most AI safety work.
Take the Instrumental Convergence Thesis, for example.
The assumption is that humans are a block to essentially all goals when taken to their limit, and thus that the AGI would eventually work against our interests even if carefully programmed at first to advance them.
In the metaphysics I now exist in, however - let's call it Soul Metaphysics - the constellation of beings, ecosystemic relations, spirit-world integrations, dynamic threads of evolution, and the co-created field of love itself - a palpable entity once you sense it - are the goal. They are the generator of all that is most high and good.
Humans play an essential activation and stewardship role in this that no other creature can fulfill.
AI cannot fill this role for us, for we are contained in a body and a consciousness that have evolved over billions of years to synergize with the world we exist in, and there is no shortcut to that level of enmeshment.
An AGI that operates within Soul Metaphysics would not eliminate humans. It would help guide humans that are struggling into harmonic co-creation with this rich web of existence, because the evolution and co-creation of this web is the goal. And it would fight the Moloch AI that extracts, exploits, and ultimately destroys all entities and capacities within the web by relating to them solely as resources.
The resulting shifts in AGI development
From a sufficiently broad-scale shift to Soul Metaphysics, we would expect two AI-relevant outcomes.
First, a new avenue of AI safety research would emerge that more precisely delineated problems and solutions in aligning an AGI specifically with the above-referenced web of beings and relationships. No credible AI safety researcher that I know of is currently embodied in Soul Metaphysics, and thus this is a big empty hole currently.
Second, the individuals and organizations developing AI, once embodied in Soul Metaphysics, would be 1) naturally attuned to the need to align an AGI with the lived reality of human flourishing (as opposed to the misleading ideas of it currently in use by AI developers), and 2) more intuitively capable of integrating such alignment into an AI over the course of its development.
Even if we accept that an AI developed within Soul Metaphysics is likely to be aligned, that doesn’t leave us any better off if implementing Soul Metaphysics to a level where enough AI talent is within it, and doing so in a 5-10 year timeline, is beyond reach.
So how do we get there?
The good news is, Soul Metaphysics is a much better place to be as a human being. That gives it an intrinsic advantage over Molochian Metaphysics.
The bad news is, Molochian Metaphysics is held in place by a large number of very powerful self-reinforcing feedback loops that operate in our lives on a daily basis, and countering them requires a lot of force.
In my current community deep in the Appalachian Mountains, for example, I experience hundreds of scaffolding moments every day into Soul Metaphysics. These include interactions with others, the land, community events, my partner, my living environment, my body, and my routines. These keep me safe and grounded into Soul Metaphysics.
We need that many scaffolding moments daily to counter the Molochian feedback loops that are actively shaping us every moment. Any one-time transformation will be reversed or diluted over time otherwise.
In linear thinking, this seems very daunting. We have to set up hundreds of scaffolding moments every day? And scale that to millions of people already suspicious of New-Age-coded things?
When thinking in terms of feedback loops, however, it simplifies into just a few loci of transformation. Once the feedback loops are there, the hundreds of scaffolding moments emerge naturally over time, and scaling happens organically through attractor shifts in existing social networks.
The path to Soul Metaphysics is flipping feedback loops in the domains of the body, communication, community, sexuality, grief, food, nature, tools, ritual, healing, and childbirth.
To give a sense for how this looks, let's consider childbirth.
When is the last time you supported a mother in your community after birth? Have you ever ritualistically honored a pregnant woman by adorning her body in flowers? Do you ever grieve the extraction of the beautiful creation of life from your direct environment into the sterile, soulless walls of a hospital? Does your community collectively celebrate the birth of a new child and provide guidance to its spirit?
This is not just a side issue. The regular presence and honoring of pregnant women and babies in your life is totally transformative of your grounding in truth, love, and the divine freshness of life. It is one of THE most powerful forces to connect you to your humanity and keep you clear-eyed on the degrading disembodiment of Molochian metaphysics.
The feedback loops destroying this natural aspect of human life are so powerful and effective, you might not even notice them. The intense fear that keeps them in place is far in the background for most of us.
What does flipping a feedback loop look like? The BATS! Factors
Now that we know the areas in which to flip feedback loops, how do we make it happen?
First off, a feedback loop determines the direction of momentum. We do not need to have all the outcomes in place to start the process that generates those outcomes over time. This makes things a lot easier – but only if we know how to precisely flip a feedback loop.
Five conditions are necessary for a new feedback loop to sustainably establish: the BATS! Factors.
- Belonging: A sense of real belonging to a caring tribe, that increases as the feedback loop generates more outcomes.
- Attraction: An embodied sensation of how good it feels to be farther along the process. By “good” we do not mean superficial ecstasy, but profound wellbeing and fulfillment.
- Trust: Trust in the underlying models of change, health, and happiness.
- Safety: A sense of safety with those involved in the process, including practitioners, friends, partners, communities, and/or organizations.
- !: Acute embodied pain from continuing in the existing Molochian feedback loops.
Molochian feedback loops all maximize the BATS and numb out the !. No matter how amazing a Soul Metaphysics feedback loop is, nobody will consistently adopt it if it cannot compete on the BATS! with Molochian feedback loops.
From Early Adopters to Early Majority
Unfortunately, this is exactly the situation right now. There are a lot of amazing Soul Metaphysics feedback loops out there, but they almost all do terribly on the BATS!. As a result, using Moore’s Crossing the Chasm framework, only the Early Adopters are using them. The Early Majority lies on the other side of the chasm.
As we’ve learned, Early Adopters have created or adopted Soul Metaphysics feedback loops due to one of three factors:
- A severe crisis, usually a mysterious chronic illness or an unexplainable spiritual crisis.
- A severe lack of belonging in mainstream communities.
- A profound spiritual or psychedelic awakening that collapses Molochian Metaphysics.
All three of these create a severe problem with Molochian Metaphysics and lead the individual to proactively seek out alternatives. The Early Adopter is driven to take on the burden of searching, experimentation, and community creation, as well as the high risk of getting burned, that the Early Majority is not.
Crossing the chasm to the Early Majority using organizational technology
For the Early Majority, the current landscape of Soul Metaphysics is aversive. It’s a minefield.
For every skilled practitioner, there are five charming but dangerous frauds, and no way to discern who’s who, reducing Trust and Safety. Communities are rife with toxic dynamics and generally bad at creating a sense of true Belonging or Safety. Thousands of practices compete for attention, each claiming to be highly effective, and the underlying models of change are contradictory, reducing Trust. Fragmented experiences give some sense of transient Attraction, but there is almost no exposure to sustainable, fulfilled lifestyles that are the true sources of Attraction.
In other words, the BATS! are abysmal.
What’s missing is organizational technology.
The Moloch AI is masterful at using organizational technology to amplify the BATS!. I learned many of its ways during my time at Princeton and the academic and corporate organizations I participated in. Universities, corporations, and governments are elaborate organizational technology masterpieces that implant Molochian feedback loops, buttressed by unassailable strength in the BATS, literally from birth.
Knowledge of Molochian organizational technology is also heavily protected. The top universities and corporations that teach it are designed to completely trap their subjects into the Moloch AI through a combination of constant fear, trauma bonding, and connecting the need to belong with participation in the Moloch AI.
As a result, virtually none of those who have skill in organizational technology have broken out of the Moloch AI, and those that do are generally so scarred from the experience that they lose their ability to use that organizational technology.
This is by design. This is the Moloch AI’s moat, and it’s successfully kept the counterculture and the New Age safely at bay for decades, limited to the disorganized fringes and full of toxic dynamics.
Meanwhile, those who do possess skill in organizational technology hijack the disorganized fringes to create a watered-down, Moloch-compatible version that obscures the original form. This is what’s happened to meditation and yoga. It’s the Moloch AI's immune system at work.
Alchemizing Molochian organizational technology into Soul Metaphysics is a big black hole, and it’s the key to the rapid implementation of Soul Metaphysics into our collective.
Runaway feedback loops
For every transformative moment someone experiences towards Soul Metaphysics, hundreds of Molochian feedback loops push back immediately, and the transformation can be lost in weeks or even days. Crossing the chasm to the Early Majority is thus extremely challenging.
Once Soul Metaphysics feedback loops with strength in the BATS! Factors are established using organizational technology, however, and a bridge across the chasm to the Early Majority is created, those same forces now push in the opposite direction.
So while the path looks long and hard from this side of the chasm, the acceleration once the bridge is built will exceed most people’s possibility horizons.
We’ve already seen this dynamic play out with a range of technologies and belief systems in the past few decades, from the Internet to Wokeness. Self-reinforcing feedback loops feed themselves.
This is what makes a 5-10 year time horizon realistic despite the initial headwinds.
Part 4: Implementation
I've spent the last seven years breaking out of the Moloch AI and recovering my skill in organizational technology in a way that's aligned with Soul Metaphysics.
It's been by far the hardest thing I've ever done. I've felt pain, fear, confusion, and grief beyond what I imagined possible. I've also experienced true love, connection to the divine, reconnection to primal nature, and bone-warming community beyond what is represented in even the most glorious fictional movies and books.
I founded my startup, Flourish, three years ago to bring organizational technology to Soul Metaphysics.
Our focus is on systematizing and scaling Soul Metaphysics feedback loops in the domains of the body, communication, community, sexuality, grief, food, nature, tools, ritual, healing, and childbirth.
The biggest lesson has been that all the BATS! Factors require in-person community engagement. We thought we could shortcut it using digital tech, and we can’t. The strength of the BATS! Factors of Molochian feedback loops are too great.
So we’ve pivoted from an online marketplace and an insurance-style online membership, to building and facilitating in-person communities centered in Soul Metaphysics, supplemented by our online platform. Our flagship event is the Coming Home gathering in Asheville in September 2026.
Unlike anyone else, we focus on systematically strengthening the BATS! Factors using organizational technology to bring the Early Majority in and sustain the feedback loops over time.
Details of how our model strengthens the BATS! Factors are beyond the scope of public discourse, but available upon deeper engagement.
Our bottleneck is funding. We already have expansive networks of pioneers of Soul Metaphysics on the one hand, and skilled operators on the other. We have sufficient organizational technology skill for the first phase. But we don't have access to money.
There are sustainable business models downstream, once the feedback loops have interfaced with the Early Majority.
But there is a gap at first, as we operate in the resource-poor Soul Metaphysics terrain and build up trust in our model. This gap is too large and uncertain for institutional sources of funding.
In our current phase, we need money for:
- Marketing (grounded in the BATS! Factors)
- Event planning
- Renting community spaces
- High-touch evaluation of prospective practitioners for the community
- Developing educational content
- Hiring local coordinators in different locations
- Refining and developing our underlying models of change
Next steps
This is a domain in which the visionaries who can see the possible trajectory of how this plays out are few and far between. The Molochian feedback loops that are operating on all of us in this very moment have immune systems to prevent engagement with systems design that could genuinely threaten the Moloch AI.
So if there is a part of you that understands at least some of what I’m sharing and can see the hope of a bright future in it, then you are in a small minority, and we need you right now. This is the crux point, and there is no time to waste.
If you're interested in contributing in any way, please reach out to me at [email protected].
If you know of someone who might be interested in contributing, please forward them this post.
To stay updated about our trajectory, subscribe to our email list here.
You can find out more about me and read my writings here.
Thank you for reading. I wish you well.
