emet-selch 🍁

emet-selch 🍁

@chaimagicks.bsky.social

↙️↙️↙️ | critic of American cultural hegemony | anti-monopoly | pro-markets & universal programs | liberal | critic of central planning (public or private) | software engineer | πŸ—οΈπŸŒπŸ§¦πŸ‡¨πŸ‡¦ πŸ•‰οΈ interests: tech, urbanism, econ, YIMBY, meditation, philosophy, occult

1.5K followers868 following1.7K posts

emet-selch 🍁 reposted

NIMBYs join forces to fight 34 social homes for homeless women and children in Maryborough. How exactly would giving them veto power improve the lives of Victorians?

You can not trust American Tech.

JoinTheATeam@kindlesnclowders.bsky.social

In Ukraine Musk caused deaths by restricting access to technology he controls. In Brazil, Zuckerberg illegally interfered in an election. And here in the US, Zuckerberg again engaged in censorship of millions of us. No Kings Not Tech Bros.

You cannot trust American Technology.

MeidasTouch@meidastouch.com

Zelenskyy says Starlink restrictions REMAIN in Ukraine, crippling interceptor drones and military operations. He says Ukraine has evidence the restrictions are DELIBERATE after testing Starlink systems in the EU, where they worked perfectly at speeds that cause connections to fail inside Ukraine.

emet-selch 🍁 reposted

The United States is actively helping Putin murder civilians in Ukraine. Trump is doing everything he can to force Ukraine to surrender to his friend and ally Putin so they can be subjugated by the Russians again.

emet-selch 🍁 reposted

This is pathetic. Ukraine tested Starlink speed restrictions in Europe and had no issues - the restrictions only exist in Ukraine. The Trump administration is actively helping Russia kill Ukrainians - and risking the lives of US and NATO advisors in Ukraine.

Alex Bozikovic@alexbozikovic.bsky.social

Perks oversaw the depopulation and gentrification of much of his ward, including displacement of many low-income people, as the city did nothing to protect tenants in small apartments and rooming houses. He spent much energy defending the planning policies that enabled these regressive changes.

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

At higher parameters, with more training and compute, it does a better job spitting out what you actually want. This means that the correct number of neurons, the correct training data, & the correct compute come together to create "better" resolution by which these machines "think" about concepts

The operative word here is: resolution. The concepts you engage with are of "higher resolution" the more you use a model like Fable or Kimi K3, and "lower resolution" when dealing with a model like Claude Sonnet. This is because Fable or Kimi K3 has a higher res representation of the concept.

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

I think what we did, technologically, is very cool. It doesn't make this technology any less wild. It makes you wonder what else you can do with neural networks. But once again, I want to be clear: neural network is the data structure being trained by the algorithm.

So from phenomenological apparence, what does a neural network like a large language model do? It takes in text (tokens) and spits out text (tokens). When there are fewer neural network parameters, it does a worse job at spitting out what you intend than if there were more neural network parameters.

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

So after all of this: what are my actual beliefs about AI? I have largely engaged with AI by thinking about what is phenomenologically apparent. Not what I believe it will be if we poured compute into it. I will explain what is phenomenologically apparent as far as I am concerned.

Basically, we took the "neural network" mathematical structure and with it grew an algorithm that uses a neural network as a data structure in a lab. We grew it by throwing at it a wide selection of our understanding the fully expressed of concepts of nature and mankind as we understand them.

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

The problem is this: Silicon Valley carved out the categories from which they will discuss AI, and those categories actually suck because of the aforementioned metaphysical commitments they take as granted, and the anti-AI people are largely operating within said categories defined by Silicon Valley

My main issue with discussing AI comes out of my just not believing in these categories. I just don't think consciousness comes out of the machine. I also don't think it will replace humans. I also don't have an opinion on data centers other than sustainability regulations instead of pure opposition

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

So, I think the problem, finally, is that the categories and lens from which VCs, Silicon Valley, and the big AI Labs choose to look at is already incorrect. Sure, they developed the cool models, but so do the Chinese, or Mistral AI in Europe, without them sounding like lunatics about it.

This brings me to my next point: VCs, Silicon Valley, and Big AI Labs have defined a cultural fight in which they are losing, because of wholesale buying into what are in essence science fiction narratives, & then fearmongering based on their fantasies, while believing that they aren't in a fantasy

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

It is this functionally unproven metaphysical belief system that drives investment patterns, especially this idea of "technocapital acceleration", and the near-religious notion of pouring money into compute in order to scale endlessly.

This is, as far as I am concerned (and I am willing to argue about it Will Stancingly), at the underlying root of the irrational exuberance that drives this tech bubble. The problem, however, is that "consciousness" (sometimes synonymous with "intelligence") is oft the realm of the most irrational

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

For some reason, we've overindexed on treating "consciousness from the machine" as rational, "evident" and "true" even if there is no actual material proof for the idea. However, this idea drives investment theses for the biggest venture capital firms, as well as the foundation LLM providers

Please sit with what I am saying for a moment: "consciousness will come from the machine" is fundamentally a metaphysical claim, because any engagement with "consciousness" whether you like it or not brings you into the realm of metaphysics

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

Because society by and large has committed to the idea of something ineffable and difficult to pinpoint as "consciousness" as one day something the machine can prove to us it is. As a result, Silicon Valley has been pouring billions into CapEX. They believe they can achieve the ineffable.

I think the words "machine god", "machine consciousness", or "singularity" are meaningless terms that verge on religious belief. They are all "ineffable" and "sublime" concepts that Western Society has constantly condemned mystics for whenever they spoke of THEIR OWN CONSCIOUSNESS

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

View full thread

I want to define myself on my own terms, and that's difficult to do. I look at AI as a technology from a very specific perspective, a perspective that much of society doesn't share because it's fundamentally having a conversation about consciousness with very bad priors

The West has the shittiest priors on the planet as they relate to the topic of "consciousness". Save for maybe the greats like Spinoza and contemporaries like Bernardo Kastrup. But they aren't treated as important thinkers by modernity, which is all in the idea that consciousness comes from material

Staying on the ball with my work as a software engineer/architect (in spite of having my own business) requires yes: talking about AI. Talking about The Good and The Bad. Being rational about it. But marketing my work means inviting in the culture war commentary layer over AI...

emet-selch 🍁@chaimagicks.bsky.social

i'm tired boss

I know I shouldn't be afraid about it, but honestly, it is terrifying given how bad the conversation about the technology has gotten. Both silicon valley and it's detractors are the worst. The reason is because they are both behaving psychotically.

Left wing and progressive notions of solidarity doesn’t actually include Indians. We are too complicated as a people. No easy narrative of oppressor <> oppressed categories to filter through.

emet-selch 🍁@chaimagicks.bsky.social

I have engaged in running defense for all sorts of groups who needed defending, but I hear the silence loud and clear when it’s Indians on the chopping block.