@IAI_TV The average human has enough difficulty being honest with itself that I’m not sure “having something at stake” is a good foundation for a theory of truth.
I don't doubt their sincerity or your experience with them. But the more consequential question may be who gets to define the problem in the first place. Who funds the framing, who benefits from delaying the distribution of AI capabilities, and who ultimately acquires the authority to decide what humanity is permitted to do with them? Good people can participate in institutional arrangements whose consequences they don't fully see. The real disagreement may lie outside the boundaries of the debate both camps have accepted.
@ericweinstein Eric, AI may reveal a different kind of mathematical genius. The hierarchy of brilliance won't disappear, but who rises to the top may change. Mathematical discovery remains the ultimate validation.
@pmarca The real UBI fight may not be redistribution vs no redistribution. It may be direct redistribution vs routed redistribution. A lot of respectable “market income” is already state mediated welfare with better paperwork.
Josh’s endorsement creates a recursive problem: he praises a warning about moral self-deception while applying it only to the other side.
Melanie’s argument is vulnerable to its own test. If moral self-deception is the danger, the warning must also apply to those defending Israel/Zionism, not only to those challenging it.
Your anesthesia work highlights how selectively consciousness can be disabled. A question I keep returning to is whether anesthesia abolishes the ability to deliberately modulate live entrainment before it abolishes entrainment itself. If that hinge fails at temporal or coherence scales not explained by network-level pharmacology, sub-neuronal physics becomes necessary rather than speculative.
If your critique of LLMs is that they’re parrots, but your counter-example is that you fine-tuned an LLM with a LoRA… then you’ve basically just sharpened the same parrot. Fine-tuning doesn’t change the architecture; it just teaches the model to repeat a different subset of the training distribution. So your experiment actually reinforces the limitation you’re pointing at, not the solution. And honestly, you’re probably far too busy to sit around coaxing an LLM into something it isn’t. Building the kind of system that actually addresses your critique is a deep, slow, architectural job; not a quick fine-tune.
@MattPirkowski Your aristocrat is the one who keeps a working feedback loop. The rest run on drift, feeding on fragments that can’t return. I read you as describing a crisis of returnability — the loss of systems that can recognize what they’ve taken in.
You’re not broken, you’re debugging a new system architecture. The baby nukes temporal continuity, so reclaim it spatially: more monitors, an externalized state, and TODO breadcrumbs. The trick isn’t uninterrupted time—it’s fast re-entry. Let the system hold the shape while you step away.
I agree with you about Harvard. Two generations raised inside social-media fields really did collapse into ideological basins, unable to see their own radicalization. Outgroups look extremist because the thought-space itself is distorted. That diagnosis is real.
Where I differ is your description of the old order as “pluralism.” For most of US history, elites weren’t plural in that sense. They shared a monocultural substrate that provided stability. Outsiders could enter, but only through merit and assimilation—competence as a gate. It was closure, not pluralism.
What we see today is not the collapse of pluralism, but the rise of something new: pluralism as overlay. Foreign elites, NGOs, and global capital flows (many of them tied back into US institutions through FRB recycling) now occupy elite slots in our hierarchy. It’s optics—a legitimacy cloak. Visibly foreign faces in leadership signal “pluralism,” while the underlying effect is adjacency without obligation.
That adjacency is perfect for exploitation. Pluralism, in this design, allows operators to extract without conscience, because there are no recursive bonds of family, community, learning, or production tying them back into responsibility for the domestic field.
The real story isn’t pluralism lost—it’s substrate breached. America’s stability came from a recursive substrate + selective assimilation gates. Those gates failed under the combined force of social-media memetic dynamics and global capital inflows. What replaced them is overlay pluralism, which dissolves reciprocity and leaves only exploitation.
Flourishing requires recursion, not overlay. Family, community, apprenticeship, and production form bonds that can rebalance slowly over time. Without that, pluralism becomes nothing more than adjacency for exploitation.
@MattPirkowski Agreed. A system that erodes consequences isn’t just failing morally—it’s following money. People do what they’re paid to do. The real question is always: who’s paying?
@ericweinstein We keep confusing Experimental experts: those who produce verifiable results you can test. And Rhetorical experts: the ones who produce consensus statements and persuasive talk.
The system rewards the latter and sidelines the former—that’s why credibility is collapsing