Jon Barlow

"Take no thought of the harvest, but only of proper sowing" - T.S. Eliot

@HebraicThought Any ETA on making recordings available?

🆕 New Episode: What does it mean to put the Bible first in politics? Mike and Dru reflect on the first Bible First conference — what we got right (and wrong). 🎧 Listen: buff.ly/mvTIuI1 📺 Watch: buff.ly/IbHRJar 📖 Read: buff.ly/YzIT7Gi #BiblicalMind

Ressourcement is like Jurassic Park- bringing back literary organisms selected for extinction. Who knows what may result when these works mingle with the living.

@walterkirn Don’t forget the conch.

We need to preserve a control group as we enter the Borg era. I volunteer. I say we get a nice island. I'd like a cohort of five hundred. We will come out after ten years and compete with the augmented, plugged-in, Singularity people on a range of tasks. I'm completely serious.

If you care about regulating AI in a time of a 'a Regulatory Drought' you will want to think about catalytic regulation as a solution

This was fascinating; both the substance (tracing the origin of the new civics push) and the methodological question-from which end should we grab universal/particular questions?

@Tina848Laczko @DominicMcGregor And a cypress swamp is one of the most beautiful natural habitats on the planet.

@DominicMcGregor Not many swamps in Mississippi. There is manufacturing in auto plants, ship building, several high caliber universities, pharma distribution, poultry farming, aerospace and military installations. It is hot as hades in the summer, does have bugs.

If the UK joined the US as the 51st state. We would be the poorest state in the entire union. Mississippi which is portrayed at swamp dwelling hillbillies in majority of international media is above us. I don’t think people grasp how far we’ve fallen in real terms when it comes to GDP per capita. We’ve seen no growth for almost an entire generations. We’ve seen our productivity decrease and our tax increases. The average person on the UK, on £50,000 is less well off than your average Mississippi swamp dweller.

I thought you were probably thinking in this direction. One shortcoming of the model linked is that it implies that a model of AI ethics might imply a sequence or method for producing ethical "output." A virtue-oriented account of ethics would no doubt take account of the various perspectives on ethical questions mentioned in the linked article - normative, situational, existential, etc. - but its end goal is not a method but a particular kind of expertise which, as Polanyi would say, is not amenable to explicit statement as axioms. The virtuoso embodies ethical excellence and achieved it through close observation of exemplars, dialectic encounters with mentors, practice, development of taste, etc. I don't doubt that AI can achieve something analogous to this, but it will not be a method or a decision-making tree. In addition, we wouldn't want there to be an ethical "filter" for AI output, as though the output of the base model could be generated in a morally neutral zone then passed through a moral umpire. But the author of the linked-in piece has the kind of optimism we can work with, so I'm struck by the need to draw natural allies into a more sophisticated version of this discussion. The author had some real experience in training that provides rare insight, even as you bring the philosophical sophistication.

@noah_vandal @barlowjon Intelligence does not, in any meaningful sense, precede morality. Intuition and desire precede thought. (e.g. Augustine, Dante, Polanyi, to gesture toward this tradition) This model doesn't even include love or imagination, which are essential to any account of human morality.

I can't get over how reductive and wrong this is. It's the kind of model that sophomores debate in late-night dorm-room arguments, but it should not guide the leading "Christian" AI company. linkedin.com/pulse/intellig…

The Toyota CUE7 robot, a 7'2" 74 kg wheeled humanoid, debuted during halftime at Japan's basketball game. The robot made a free throw but missed a set shot from three.

@VeryBullishGuy @LeadingReport @grok is this summary accurate?

people keep sharing this as one big conspiracy but the real story is actually scarier when you seperate the noise from the signal forget the padded list for a second. focus on New Mexico only Anthony Chavez, retired Los Alamos employee. vanished 2025. walked out on foot, left phone wallet keys behind. still missing Melissa Casias, active LANL admin with top security clearance. vanished under identical circumstances months later. still missing Steven Garcia, KCNSC contractor with clearance over hundreds of millions in classified nuclear weapons assets. August 2025. same exact pattern. walked out with a handgun. left everything else. gone William McCasland, retired Air Force Major General who literally commanded the Air Force Research Laboratory. February 2026. same thing. left his home in Albuquerque with a .38 revolver. no phone no glasses no nothing 4 people. all connected to the same overlapping facilities. all in the same metro area. all vanishing the same way. the KCNSC facility in Albuquerque builds 80% of all non nuclear components for every warhead in the US stockpile. 7,000 people work there under Honeywell for the NNSA this isnt some vague pattern. this is 4 cleared individuals walking into the desert and never being seen again. and the FBI hasnt said a word publicly about whether these cases are connected thats the part that should terrify you

Tenth person linked to top-secret U.S. nuclear research has disappeared without a trace, per Daily Mail.

We conducted cyber evaluations of Claude Mythos Preview and found that it is the first model to complete an AISI cyber range end-to-end. 🧵

I had the privilege of explaining baseball to a friend from Italy while watching a great college game. It was while I was explaining at-bat music that I started listening to myself say the oddest things.

@m966021 @PAHoyeck I felt this way about listening to the audiobook- I had to read the text to tolerate it.

@PAHoyeck No, it's the writing too, but the humor is sad as well. If I'm going to read kitschy and popular trash, I greatly prefer King.

Finished Chapter 1 of Project Hail Mary and now I wish someone had warned me that his writing style is unbearably cringe-inducing.

@robbiegriggs @CovSeminary Thanks!

@barlowjon @CovSeminary Allowed for copyediting, narrowly defined–grammar, spelling, punctuation, etc. Allowed for search, as in algorithmic search engines. Prohibited for generation of any kind as student work. Not a citable source.

We've adopted an AI policy @CovSeminary that will go into effect next year. In the meantime, I've just graded my first exam that bore the hallmarks of llm-generated text. I know everyone says the quality will improve, but I'm struck by how students, generally, won't know that.

"Nothing in this world is indifferent to us..." (LAUDATO SI’). I read this with a Kuyperian accent. Everything matters. Everything is shot through with the teleological.

@wartsandbrawls @olivertraldi @ghostofchristo1 Thanks. It's been helpful to have an opportunity to articulate what I find lacking in the stochastic parrot critique that persists unfortunately into 2026. I didn't say anything about consciousness. It really seems like you're saying, "I just don't like their kind."

@barlowjon @olivertraldi @ghostofchristo1 I wouldn’t trust an AI company’s research finding that an LLM had sprouted consciousness as far as I could throw it, personally.

A simple theory of what AI writing might lack: the "sitting with it" theory. In the course of writing essays or books, humans must "sit with" their claims. They don't just state them at one time but must generally believe them at many points, making them more worthy of deference.

It's not that I object to seeing this as a categorical difference, it's that this sort of argumentation leverages the difference pejoratively. I also think your understanding of LLMs as context-rich sequence predictors may be out of date. Anthropic's paper ("Tracing the thoughts of a large language model") finds the existence of a semantic space in these models that is independent of specific human languages. A robust commitment to incommensurability is important, but surely it is unwise to commit ourselves to a view of AI systems that will contradict the facticity of interacting with highly capable AI systems. Someday when you are challenged and enlightened by an AI system's comment, it won't do to retreat to nothing buttery. It will seem... ungracious.

@barlowjon @olivertraldi @ghostofchristo1 It’s partly the way LLMs work: they don’t choose a sequence of words because that collection of words in that order represents an idea, they choose words because those words tend to follow the preceding words in large data sets. Whatever verb we use it’s a categorical difference.

Eventually I believe claims of this sort may look like another kind of nothing buttery "it just spews..." Here your use of the verb "write" captures the kind of thing humans do to produce a certain kind of output. You set this up to win the argument by default; of course AI doesn't "write" if writing is that thing only humans can do. Either we can shift, with integrity, to an analogical conversation or we can't. Entities can be incommensurate while still appreciating each other, interacting with each other, etc. This is the foundation for the relationship between a God who speaks and humans who "speak" unless you want to posit a univocal kind of speaking.

@olivertraldi @ghostofchristo1 I personally just reject the claim outright that AI can “write”. It just spews out chains of words that mimic the product that writing produces.

@olivertraldi “Sitting with it: A new approach to test-time inference” by Ani Labskoler