Why I Bet on Alignment

The downbeat at Radio City lands in your sternum before you hear it. You feel the room first — six thousand bodies leaning forward in the dark, the rake of the seats, the smell of stage dust and hot lamp gels. Then the kick drum, then the bass, and the whole hall goes from waiting to moving in the span of a single eighth note. I am thirty feet stage left, five-string in my hands, in-ears tuned to the click and the cue mix. The conductor glances over. I nod. We are inside the song now and the song is inside the room and somewhere between the third bar and the fourth I notice, again, that this is the easiest thing I have ever done.

That sentence is going to sound obnoxious no matter how I frame it. Let me try anyway.

I was, for a window from roughly 2015 to 2020, the best electric bass player in the world. Radio City many times over. Broadway pits. Elite orchestras. Sessions in the studios you would name if I asked you to name studios, with the musicians whose names are printed on the records that taught me to play. Every dream I had at fifteen came true between twenty-five and thirty. I am not bragging. I am setting up a confession.

The confession is this: I was the best in the world at a very easy, uncontested thing. Bass is technically demanding. Bass at the highest level requires a thousand hours on fundamentals and a thousand more on feel, ear, and reading. That is not what I mean. What I mean is that the space of bass is small. There are perhaps a few hundred people on earth seriously competing for the chair I held, and most of them do not actually want it as badly as I did. Once you are inside the top room there is no further depth available. There is only repetition with more decoration. You play the song again. You play it slightly better. You play it on a slightly bigger stage. You die.

I do not say this to be ungrateful. The bass years made me — they taught me how to commit to a craft, how to listen, how to be the steady center of a thing larger than myself. I love bass. I will play bass until I am dead. But somewhere in 2019 I started waking up with a feeling I could not name, and the feeling, when I finally translated it, said something like: this game is over and you have won it and the prize is too small.

So I went looking for a harder game. The early-2020s gaming boom was the obvious move — millions of competitors, real ranking systems, a contested arena where the best of the best could be measured against each other in something close to real time. I have always loved competition. I have always loved video games. I figured if the bass world was a pond, esports was an ocean, and I would find the depth I was missing inside the saltwater.

I climbed. League of Legends to a rank I am still proud of. Valorant to a similar place. Fortnite, briefly, into the lobbies where the sponsored kids hang out. The games were hard. The competition was real. The hours were long in a way bass had never demanded. After two years of it I was profoundly, baffled-by-it depressed. It made no sense. I was playing the games I loved, all day, at a level I had dreamed of as a teenager. I should have been thrilled.

Instead I felt myself disappearing into pixels. I was not moving my body enough. I was not contributing enough to society or to the universe or to anything I could point at without flinching. The stakes were artificial in a way I could not unsee once I had seen it. Every match was loud and contested and intense, and at the end of it nothing in the world was different. Nobody's child ate. Nobody's road was paved. Nobody's mother's tumor shrank. The intensity was real but it was not connected to anything. It was strength training that produced no work. After a while the body knows.

I have a constitutional rule for myself now, written down, that says I will not confuse stimulation with aliveness. I learned that rule the hard way, sitting in a dark room at two in the morning with a headset on, having just won, feeling nothing.

The pivot to software was not romantic. It was a clean-eyed assessment of what I am for. Ken Wilber, in A Brief History of Everything, defines goodness as the forward evolution of consciousness — the tendency of the universe to wake up to itself, in deeper and wider patterns, across organisms and minds and cultures and whatever comes next. I read that line and something inside me settled. It gave me a unit of measure. A life is good to the degree it generates that forward motion. A career is good to the degree the work is leveraged on it.

Once you have that frame, the question of where to put your years becomes almost embarrassingly tractable. What is the lever with the largest expected effect on consciousness over the coming century? Not music. Not games. Not anything I had previously considered. The honest answer, looked at squarely, is artificial intelligence. The systems being built right now will either extend the trajectory of waking up — more minds, deeper minds, wider circles of care — or they will flatten it into a hall of mirrors. Either way the lever is enormous. Either way the leverage is going to be pulled.

So I got a software job. Interactive Aptitude — a small DoD contractor funded mostly by SBIR grants. I write Python. I write TypeScript. I work with model code and pipeline code and the unglamorous infrastructure on which actual research depends. It is the apprenticeship phase. I am not running the lab yet. I am learning to run the lab. The constitution I wrote for myself has a clause on this: I shall not waste my best years on work I know to be trivial, corrosive, or misaligned. Every line of code I have shipped in the last two years has been auditioning, internally, against that clause.

The natural follow-up question is why alignment specifically and not, say, capability. Capability is the loud lane. Capability is where the money is, the prestige is, the timeline-defining demos are. If I wanted to be near the action without getting my hands dirty on the hard problem, capability is the obvious place to set up shop.

Capability is also being taken care of. There are tens of thousands of brilliant, well-resourced, well-incentivized people racing to make these systems more powerful. They do not need me. The capability frontier does not move faster because one more competent generalist shows up to push on it. It moves at roughly the rate it would have moved without me, and I would be a rounding error inside the larger curve.

Alignment is different. Alignment is the part of the work where consciousness either gets honored or gets steamrolled. It is the slower, smaller, less-funded, less-glamorous lane, and an additional serious person actually changes the slope of the curve in that lane. The marginal value of one more committed researcher in alignment is many orders of magnitude higher than the marginal value of one more committed researcher in capability. This is not a sentimental claim. It is an arithmetic one.

And the stakes are asymmetric in the way that matters. If we get capability right and alignment wrong, we have built a very fast machine for producing outcomes nobody wanted, including the people who built the machine. If we get alignment right and capability is a little slower than it might have been, we have lost some quarters of GDP growth and gained a future. The expected-value calculation here is not subtle. The bet that pays the universe back, if it pays at all, is the alignment bet.

That is what I am betting on. Not because I am certain it works. Not because I have some private oracle that tells me the technical problem is solvable on a timeline that matters. I am betting on it because the bet is the only one whose payoff structure is consistent with what I think a life is for.

The constitution I keep has a section on artificial intelligence. The relevant clause says that because AI may become one of the most powerful forces in history, I am to approach it with seriousness, imagination, and moral courage — that my work with it is to be directed at alignment with the deepest good available to us, not merely convenience, not merely profit, not merely obedience. I wrote that down before I was ready to live up to it. The point of writing things down is to hold yourself accountable to a self you have not yet become.

What I can tell you, from inside the apprenticeship phase, is that the work feels different from the bass work and different from the games. There is no ceiling. The space is enormous. The contest is real. The stakes are connected to consequences in the world. My body is moving more, because I am no longer chained to a chair under stage lights or under monitor lights but living a life where I write code in the morning, lift in the afternoon, and chase a two-year-old across the yard at five. The depression that haunted the gaming years is gone. The satisfaction that haunted the bass years — the I have arrived and there is nowhere left to go satisfaction — is also gone, and good riddance. There is, in fact, more left to go than I will live to finish.

That is the bet. A musician who hit the ceiling of an easy game. A gamer who hit the floor of a hollow one. A software developer in an upstate New York office, training toward founding a lab. A father of two daughters whose lives will overlap with whatever this technology becomes. A husband of a woman who has watched all three pivots and has not flinched. The bet is that the work I am preparing to do is the highest-leverage work available to a person who takes consciousness seriously, and that the only honest response to standing in front of the lever is to pull.

I do not know if it works. I am confident enough to put a life on it. That is the only kind of confidence I trust anymore.