As an ex-prefacer, imo it’s a hedge against the nudging recognition that the thought being hedged is based on a pyramid of iffy connections (usually more word-of-mouth type “science”), rather than an articulable, concrete a->b->c.
My interpretation is “Programming in their nature” -> their being, meaning the molecular structure that defines them, is a program, a complex program, one executed by the interpreter that is physics.
You might be interested in Catholic theology + philosophy. It takes a LOT of background, and a lot of time to accumulate it, but you clearly have some already, and once you get comfortable with reframing terms that have a lot of societal baggage attached to simpler, honest understandings (for example, God, as a sky dude vs the universe as a mind), you can go so far.
> How much compute could you really pack into a Dyson swarm if that was your only objective? what are the ultimate theoretical bounds for compute density/efficiency?
The latter concept is known as the Boltzmann limit, and if you’re interested in these types of questions, I highly recommend checking out Isaac Arthur’s youtube channel and backlog[1].
My interpretation was that they’re referring to legal/financial engineers in the context of corporate butheshareholderism, where rather than trying to e.g. build a good nation, the goal is to maximize profits.
It would be interesting to see if a coding LLM trained for tool calling like GLM 5.3 would work well as a Jev model, or maybe even a flow where a model generates options (eg for a plan) and then uses a Jev to refine/optimize the path. Or similarly where else in a harness they’d help.
“Know” isn’t a perfectly useful word here. The mind still experiences sans material stimulus, for example dreaming. And it experiences in a progressively less goal/immediate-task-relevant-stimulus manner as stimuli go down (default mode network activity, seen in dreaming, sensory deprivation etc) - i.e. the mind experiences, and processes (dreams about taxes, and so on) independent of the world.
What support? Asahi had to fight to exist, especially early on when they essentially had to reimplement much of the graphics pipeline. And now its in mainline linux - short of a complete hostile lockdown of the bootloader, it’s not like they support other OS’s very much as is.
Who said anything about agentic, or reasoning? Those are post-hoc additions to autoregressive usage of LLMs. The base case of predicting next tokens generally (rather than with a specific objective) requires a faithful representation of the input embeddings, which is plausibly repurposed for classification.
Essentially because attention introduced a way to scale un/self-supervised learning to the level of data out there, and learnable inference time 0-shot feature selection. Impressively in a autoregressive, unidirectional manner.
Stories and music, their character, comes from combined motifs and themes. These are what gives them meaning. Creation is tied to meaning; one can’t separate human meaning out of the conversation. Without human meaning - new perspectives gained, beauty found - all things are equivalently pointless, and all an LLM is doing is shifting (very high-order, having gone through the steps to end up in training data) entropy around.
A good router on the user-facing end would be nice, but I’d rather be able to get a feel for which car I’m driving and pick one depending on the task, than have to hope the rental agency knows what I need.
Language is absolutely pattern matching, and the whole thing is that the more patterns you put in, the more the transformer can pull out what’s useful. Agents are this effective because people figured out that training on tasks being completed gets the models to complete tasks - that’s pattern matching.
And it can only go so far, where those patterns are feasibly captured, eg internet discussions and coding and mathematics. It won’t measure up to the scaling laws of the physical world.