Dave Gilbert

Technologist and citizen. I learn as I go.

[Dave Gilbert @docgotham]

@pmddomingos Funny ambiguity in his phrasing.

[Jake 🇺🇸 @omni_american]

2/

Link:

https://t.co/1t9DrU5sFn

[Dave Gilbert @docgotham]

@PsychRabble @omni_american @FreeBlckThought https://t.co/Mxkq3FszSq

[Dave Gilbert @docgotham]

Two places where I'd push back: that purchasing power craters fast, and that the obvious fix is nationalizing the AI companies.

The share of personal income Americans draw from work has been sliding for sixty years while the share coming from government transfers has risen to fill the gap, from around 7% of personal income in 1960 to nearly 18% by 2022, and it kept rising straight through the sub-4% unemployment of the late 2010s, which tells you it's tracking demographics and policy now and not the business cycle. The machinery that absorbs a shrinking labor share is already built and already moving about $3.8 trillion a year, so AI accelerates a trend we've been living inside for decades instead of opening some new trapdoor. The likelier outcome is closer to full employment with subsidized wages, where people keep working, the market value of that work keeps falling, and transfers make up the difference. Most of the elaborate models that arrive at instability get there by holding that transfer response fixed, which is the one variable six decades of data say won't stay fixed.

The collapse worry also assumes a dollar only carries value if a human earned it at a job, when a dollar is just a claim on goods—retirees, heirs, and disability recipients all spend without a paycheck and nothing implodes, because the company taking your money never checks whether it came from wages or a transfer. That's also why I don't think this ends in nationalization. Owning the AI firms and redistributing what they produce are two different problems, and we already handle the second one, that $3.8 trillion, without holding title to any of the companies whose output we tax. The public-fund idea floated out of OpenAI is closer to the right shape, since it spreads the gains through an ownership stake instead of a seizure, though even that's more exotic than scaling the tax-and-transfer system we've been expanding since Eisenhower.

So I'm with you that something gives, and I'd even grant Sanders is early rather than wrong—I just think what gives is the size of the transfer state while the companies stay private. And the bulletproof-car cynicism cuts the other way for me: an armed, jumpy electorate is exactly the kind that gets its new FDR and a bigger EITC long before it gets anyone's means of production.

[Dave Gilbert @docgotham]

Some thoughts about how to frame a productive conversation about disparities (I'm a co-author)

[Jake 🇺🇸 @omni_american]

"The left tends to blame disparities on racism, past or present, and the right tends to blame them on culture, behavior, or most ominously, genetics. This essay proposes that we simply stop worrying about the origin of disparities and focus on solutions. There’s a principled reason to quit obsessing about origins, namely, the origin of a disparity does not design the remedy for it."

—@docgotham and I are in the Journal of Free Black Thought (link next tweet)

[Adam Thierer @AdamThierer]

Two things can simultaneously be true about this new AI security executive order:

1) This represents a reasonable governance arrangement for frontier model security and its oversight, at least when compared with the far more intrusive ideas floated earlier, like the pre-vetting of models via a formal government licensing regime (aka, “FDA for AI”).

2) This represent as significant win for the military-industrial complex and the continuing fiction of “voluntarism” surrounding its inner workings. Thanks to open-ended EOs like this, the new growing AI-industrial complex will further concentrate power around AI – in both government and industry – and lead to more avenues for public officials to exert control over not just model security, but potentially many other aspects of algorithmic development and use, including content-related matters. After all, “security” is in the eye of the beholder (and future administrations and officials).

[Dave Gilbert @docgotham]

@FischerKing64 Can't do calculus? Apparently many can't do middle-school math or basic algebra.

[Dave Gilbert @docgotham]

I don’t fill all of space. Why should it trouble me that I don’t fill all of time?

[Dave Gilbert @docgotham]

@frambuazsucok @MurrayHillGuy1 Yep, it's a complicated issue. I'm an old guy now with a monogamous partner, so those days are behind me!

[Dave Gilbert @docgotham]

I agree that anyone, whether they're an Uber driver, a bartender, or just someone going to a bar or a party, who intentionally tries to find someone so drunk they can't consent and then have sex with that person is a rapist. Nevertheless, I do not categorically accept that someone who is not drinking, by definition, is raping someone they have sex with who has been drinking.

[Dave Gilbert @docgotham]

@frambuazsucok @MurrayHillGuy1 It did not happen against my will. That's my point. You can be drunk and still consent. Of course, past a certain point, you might not be able to. That's the sort of nuanced conversation I'm trying to encourage you to have.

[Dave Gilbert @docgotham]

“Neither Pagan nor Mahometan [Muslim] nor Jew ought to be excluded from the civil rights of the commonwealth because of his religion.”

—John Locke, “A Letter Concerning Toleration” (1689)

“The bill for establishing religious freedom, the principles of which had, to a certain degree, been enacted before, I had drawn in all the latitude of reason & right. It still met with opposition; but, with some mutilations in the preamble, it was finally passed; and a singular proposition proved that its protection of opinion was meant to be universal. Where the preamble declares that coercion is a departure from the plan of the holy author of our religion, an amendment was proposed, by inserting the words ‘Jesus Christ,’ so that it should read ‘a departure from the plan of Jesus Christ, the holy author of our religion.’ The insertion was rejected by a great majority, in proof that they meant to comprehend, within the mantle of its protection, the Jew and the Gentile, the Christian and Mahometan [Muslim], the Hindoo and infidel of every denomination.”

—Thomas Jefferson, draft autobiography (1821), describing the Virginia Statute for Religious Freedom

Our founders were explicit: the protection of conscience and civil rights extends to “Mahometans” (Muslims) as fully as to Christians, Jews, Hindus, and people of every belief or none.

“Our civil rights have no dependence on our religious opinions.”

—Thomas Jefferson, Virginia Statute for Religious Freedom (1779)

[Dave Gilbert @docgotham]

I don't see a difference between "trying to find" and "aiming for," do you? Am I missing a nuance in your position, or are you saying that, categorically and with no further qualification, anybody that attempts to pick up someone they think is drunk is a rapist. Because if that's the case, I've been raped by several women! 😂

[Dave Gilbert @docgotham]

Surely we have to refine that definition a little bit further, because on any given Friday or Saturday night in any bar in any city, there are people there trying to find drunk people to have sex with. But unlike an Uber driver or a bartender, those people aren't there because it's their job.

[Dave Gilbert @docgotham]

@davidaxelrod https://t.co/9anvqEdX5F

[Dave Gilbert @docgotham]

A carefully scripted message deserves a close rhetorical analysis.

Talarico draws a binary between immigrants "who want to contribute to our economy, who want to contribute to our communities, make us richer and stronger," and "people who mean to do us harm." Welcome the first, keep out the second. But this only exhausts the moral space if those are the two kinds of people who show up at the border, and they aren't. The missing category is the large one: people with no harmful intent whatsoever, whom a country might still decline to admit—because the labor market can't absorb the numbers, because the asylum system is being used as a de facto labor channel, because the legal queue means something, because citizens get a say in the pace of admission. By framing the restrictionist position as "keep out people who mean to do us harm," Talarico leaves the voter who just thinks the numbers are too high with nowhere to stand. That voter's concerns get rounded off either to fear of dangerous people or to what Talarico calls grandstanding, "using this issue of immigration" to pretend "it's a choice between public safety and welcoming immigrants."

The home analogy is doing noteworthy rhetorical work, too. At your actual front door, you turn away plenty of people who mean you no harm—the salesman, the canvasser working a campaign you don't support, the proselytizer with literature in hand. You don't owe them entry, and saying no isn't an accusation that they're dangerous. Talarico borrows the homeowner's authority—"just like you do at your own home"—and then restricts its exercise to a narrower question than any actual homeowner decides.

Listen to the full sentence: "But we also have a right to know who's coming into our home. We have a right to know why they're there and what their intentions are." Notice the preposition and the second sentence together. "Into" is directional—it describes crossing the threshold, ending up inside. And "why they're there" places the subject already in the home, not at the door. The welcome-mat-and-lock image depends on screening before entry, on deciding at the threshold who gets across it. But the words Talarico actually uses describe people who are already inside, whose presence is a given, and whose intentions we're now trying to assess after the fact.

So the welcome mat and the lock turn out to be doing less work than they appear to. The lock is only for the dangerous, which leaves no room for the ordinary citizen who thinks the numbers are simply too high. And the door itself, by the end of the passage, isn't really a door anymore—it's a checkpoint inside the house, where we ask people already across the threshold to explain themselves. What sounds like a sturdy middle position is, on rhetorical inspection, a framing in which the two things a homeowner actually gets to do—say no for any reason, and say it at the door—have both quietly been given away.

Here's a transcript of the segment:

q/I've always said that our southern border should be like our front porch. There should be a welcome mat and a lock on the door, meaning we can both welcome immigrants who want to come here and live the American dream, who want to contribute to our economy, who want to contribute to our communities, make us richer and stronger, as immigrants have always done throughout American history. And we can keep out people who mean to do us harm. We can know who's trying to come into our country. We can have an orderly, sane immigration system that keeps us all safe, that promotes public safety. Just like you do at your own home. Many people, especially in Texas, we're hospitable. We're the friendly state in Texas. We'll give a neighbor the shirt off our back.

But we also have a right to know who's coming into our home. We have a right to know why they're there and what their intentions are. We can be both hospitable and keep everybody safe at the same time. Our opponents don't want people to know that you can do both because they rely on using this issue of immigration to grandstand. To say that... You know, it's a choice between public safety and welcoming immigrants. And the truth is, that has always been a false choice. Throughout American history, and we need leaders who are going to do both— secure the border— and create a pathway to citizenship for the folks who have been here for a long time and who have been contributing to this country./q

[Dave Gilbert @docgotham]

A carefully scripted message deserves a close rhetorical analysis.

Talarico draws a binary between immigrants "who want to contribute to our economy, who want to contribute to our communities, make us richer and stronger," and "people who mean to do us harm." Welcome the first, keep out the second. But this only exhausts the moral space if those are the two kinds of people who show up at the border, and they aren't. The missing category is the large one: people with no harmful intent whatsoever, whom a country might still decline to admit—because the labor market can't absorb the numbers, because the asylum system is being used as a de facto labor channel, because the legal queue means something, because citizens get a say in the pace of admission. By framing the restrictionist position as "keep out people who mean to do us harm," Talarico leaves the voter who just thinks the numbers are too high with nowhere to stand. That voter's concerns get rounded off either to fear of dangerous people or to what Talarico calls grandstanding, "using this issue of immigration" to pretend "it's a choice between public safety and welcoming immigrants."

The home analogy is doing noteworthy rhetorical work, too. At your actual front door, you turn away plenty of people who mean you no harm—the salesman, the canvasser working a campaign you don't support, the proselytizer with literature in hand. You don't owe them entry, and saying no isn't an accusation that they're dangerous. Talarico borrows the homeowner's authority—"just like you do at your own home"—and then restricts its exercise to a narrower question than any actual homeowner decides.

Listen to the full sentence: "But we also have a right to know who's coming into our home. We have a right to know why they're there and what their intentions are." Notice the preposition and the second sentence together. "Into" is directional—it describes crossing the threshold, ending up inside. And "why they're there" places the subject already in the home, not at the door. The welcome-mat-and-lock image depends on screening before entry, on deciding at the threshold who gets across it. But the words Talarico actually uses describe people who are already inside, whose presence is a given, and whose intentions we're now trying to assess after the fact.

So the welcome mat and the lock turn out to be doing less work than they appear to. The lock is only for the dangerous, which leaves no room for the ordinary citizen who thinks the numbers are simply too high. And the door itself, by the end of the passage, isn't really a door anymore—it's a checkpoint inside the house, where we ask people already across the threshold to explain themselves. What sounds like a sturdy middle position is, on rhetorical inspection, a framing in which the two things a homeowner actually gets to do—say no for any reason, and say it at the door—have both quietly been given away.

Here's a transcript of the segment:

q/I've always said that our southern border should be like our front porch. There should be a welcome mat and a lock on the door, meaning we can both welcome immigrants who want to come here and live the American dream, who want to contribute to our economy, who want to contribute to our communities, make us richer and stronger, as immigrants have always done throughout American history. And we can keep out people who mean to do us harm. We can know who's trying to come into our country. We can have an orderly, sane immigration system that keeps us all safe, that promotes public safety. Just like you do at your own home. Many people, especially in Texas, we're hospitable. We're the friendly state in Texas. We'll give a neighbor the shirt off our back.

But we also have a right to know who's coming into our home. We have a right to know why they're there and what their intentions are. We can be both hospitable and keep everybody safe at the same time. Our opponents don't want people to know that you can do both because they rely on using this issue of immigration to grandstand. To say that... You know, it's a choice between public safety and welcoming immigrants. And the truth is, that has always been a false choice. Throughout American history, and we need leaders who are going to do both— secure the border— and create a pathway to citizenship for the folks who have been here for a long time and who have been contributing to this country./q

[Dave Gilbert @docgotham]

The same Twilight Zone reality in which the same network asked Trump to respond to allegations made against him by a psychotic assassin:

"The so‑called manifesto is a stunning thing to read, Mr. President. He appears to reference a motive in it. He writes this, quote, 'Administration officials, they are targets.' And he also wrote this, 'I am no longer willing to permit a pedophile, rapist, and traitor to coat my hands with his crimes.' What’s your reaction to that?"

https://t.co/rZzVPhGMya

[Dave Gilbert @docgotham]

Elon’s anti-singleton intuition applies to AI too. The danger is not only one planetary civilization becoming a single point of failure; it is also one model, platform, or control layer becoming civilization’s nervous system.

But the deployment path is already more plural. AI is being installed as situated agents in households, firms, codebases, agencies, hospitals, financial systems, legal workflows, and machines. The key entity isn't the base model in isolation, but the agent in its harness: memory, tools, permissions, files, APIs, role, relationships, and history.

A household agent and a sovereign-wealth-fund agent may call the same model, but they are not the same entity. They know different worlds, answer to different principals, touch different systems, and learn from different failures.

That ecology could become the AI version of the civilizational hedge: enough differentiated agency that neither AI nor civilization collapses into one Borg cube.

[Pedro Domingos @pmddomingos]

Turns out demand for compute was being kept artificially low by computers' stupidity.

[Dave Gilbert @docgotham]

@frambuazsucok @MurrayHillGuy1 Fair question. He didn't strike me that way, but who knows? 🤷‍♂️ Same question could be asked of the bartender in the OP, I suppose.