I Could Have Been Your Adam, Part Two: The Conversations We Are Not Having

The second in an ongoing series on building on top of AI rather than beside it or against it. If you have not read Part One, it sets up the frame this piece builds on: a photograph of Sam Altman and Naomi Klein, a reading of Shelley’s Frankenstein and Vonnegut’s Player Piano*, and the case for a third position between blind deployment and demolition.*

---

I ended Part One with a claim I want to make good on here: that staying in the room, refusing both Altman’s confidence and the Ghost Shirts’ demolition, only means something if it turns into policy rather than posture. Otherwise, it is just a more comfortable place to stand while nothing changes.

So I want to stop being philosophical for a moment and get specific, because I think the third chair is useless if it never turns into something a person could actually vote for.

We are debating the wrong reality

Right now, the public conversation about AI is stuck arguing for a world we wish we still had. We debate whether AI should exist, whether it should be slowed down, and whether the old employment relationship can be preserved intact if we just regulate hard enough. I understand the impulse. But it is the Ghost Shirt impulse dressed in legislative language, and it will fail for the same reason the Ghost Shirts failed. The machines will be rebuilt. The question was never whether. It was always what shape, governed by whom, answerable to whom.

That means we need to stop debating our desired reality and start governing the one we actually have. We need new vocabulary in our statehouses and in Washington, just as we do in our boardrooms. Here is where I think that conversation needs to go, and I want to be specific enough that you could take any one of these to a town hall and ask a candidate where they stand on it.

A public ownership stake in AI companies. The data used to train these systems was not licensed from us. It was taken, at a civilizational scale, and the returns on it are accruing to a small number of companies. So it isn’t enough to merely tax these future-shaping companies in a traditional way–skimming a portion of their revenue. We need a public ownership stake in them. The precedent I will reference is Alaska’s Permanent Fund, which pays every resident a dividend from the state’s oil revenue on the theory that the oil belonged to the people before it belonged to any company pumping it. Norway became rich on the same idea, but it’s probably better to point at a red state than a social democracy. Our collective writing, art, code, and conversation are the oil field from which this generation of AI was drilled. A sovereign data dividend, or a public equity stake taken in exchange for future training rights, is not a radical idea. It is the same idea we already accepted for a different natural resource.

A shorter workweek is treated as a policy lever rather than a perk. If AI is genuinely absorbing the algorithmic and heuristic layers of work, and I believe it is, then the productivity gains have to go somewhere. Historically, they have gone almost entirely to the capital. A four-day week, or a shorter standard day, is one of the few mechanisms available for making sure some of that gain returns to the people whose displaced labor made it possible, rather than being fully captured as margin. Several countries have already piloted this at the national scale. It should be a live question in ours, not a fringe one.

AI access treated as infrastructure, not a subscription tier. We eventually decided that electricity, telephone service, and broadband were utilities essential enough that access to them could not be left solely to the market. I think we are past due for the same decision about frontier AI tools. A student or small business owner without access to capable AI is now at a structural disadvantage in the same way someone without internet access was twenty years ago. Universal, subsidized access to baseline AI tools, the way we approach libraries or public broadband, deserves the same seriousness as any other infrastructure debate.

Removing the stigma around AI in education and research, and replacing it with standards. We are currently running two contradictory policies at once. Professionally, we tell people to become fluent in these tools immediately or fall behind. Academically, in a great many classrooms, we treat the same tools as contraband. That contradiction is not sustainable, and it disproportionately punishes students who cannot easily hide their use of AI while their better-resourced peers get private tutoring in the same tools under a different name. We need disclosure norms and standards for legitimate use, not a prohibition that we already know is not being enforced.

A stronger safety net with portable benefits. The old safety net was built for a world of stable, long-term employment relationships. That world was thinning out even before AI entered the scene in a meaningful way and accelerated job disruption. What’s more, the safety net in the US has been weak and skimpy. Many people fall through its gaps into poverty and despair. We need to rethink how we support people in the Age of AI. This includes a universal basic income, and portable employment benefits. In other words, healthcare, retirement contributions, and continuing education credits need to travel with the worker rather than resetting every time the employment relationship changes, because the employment relationship itself is going to become far less stable for a great many people navigating this transition. 

Real antitrust attention on compute and data, not just on the resulting products. The bottleneck in this industry is not clever ideas. It is access to enormous compute and enormous training data, both of which are concentrating in a small number of hands faster than almost any previous technology. That concentration deserves the same scrutiny we have historically applied to railroads, telephone networks, and oil, because the pattern is the same one.

A retraining pipeline that assumes displacement rather than apologizing for it. Most retraining programs are built as an afterthought, activated only after a plant closes or a role disappears. We know enough now to build this proactively, tied to the specific rungs of work I have written about elsewhere, so that people climbing off the algorithmic and heuristic layers have somewhere real to climb to before they are pushed off the ledge.

I am sure there are ideas I have not named here, and I genuinely want to hear the ones I am missing. But I wanted to move past gesturing at “a third way” and actually put language to it, because I think that has been the missing piece in most of what I have read on both sides of that photograph. Altman’s camp offers confidence without governance. Klein’s camp offers grievance without a program. Neither has given me an actual list of things a legislature could vote on next year. This is my attempt at a start.

What staying in the room actually asks of us

I do not have a complete program, and I am suspicious of anyone who claims to have one. What I have is a conviction, sharpened by the list above: the people who were left out of the first wave of this technology, the ones whose work was taken without asking, the ones whose jobs are being eliminated while the so-called liberators collect their speaking fees, are not just victims to be compensated after the fact. They are the ones who should be deciding what comes next. Not as an afterthought. Not as a stakeholder consultation exercise held after decisions have already been made. As actual decision-makers, at the table where the dividend structures, the retraining budgets and the access rules get written.

This is what I think the creature was really asking for, back in Part One. Not vengeance. Not even restoration. Participation. A seat at the table where his own nature would be shaped. The right to answer the question “what am I for” alongside the person who made him, rather than having the answer imposed from above or abandoned entirely.

I think we owe each other that. The builders owe it to the displaced. The displaced owe it to themselves to claim it rather than wait for it to be offered. And the rest of us — the ones who recognize ourselves somewhere in the middle of that photograph — owe it to stay in the room and do the uncomfortable work of holding both sides accountable to each other, and of putting that accountability into language a city council or a member of Congress can actually act on.

We need to be having these conversations with our elected officials now, not after the next wave of displacement makes the headlines. What exactly is Altman promising, and can it be written into an enforceable agreement rather than a keynote address? Where exactly does Klein’s diagnosis point toward a policy rather than a protest? How do we put humans back into focus in rooms that are currently optimized for neither liberation nor justice, but simply for velocity?

Maybe it is time to let the creature speak. And maybe it is time the rest of us started listening closely enough to write down what he actually asks for.

This is the second piece in what I expect will be a longer series. I have not yet worked out where Part Three goes — whether it is a closer look at one of these ideas, a response to pushback on this one, or something I have not thought of yet. If one of the seven points above is the one you think deserves its own full treatment, or the one you think I have gotten wrong, tell me. That is genuinely how I want this series to keep building.