Eryk Salvaggio

How is Sam Altman's World infrastructuring a post-democratic AI future? In this paper published in AI & Society, @laura4lano.bsky.social and I analyse how World's value is built on sociotechnical fictions & existential risk narratives that advance a cyberlibertarian transition. tinyurl.com/c8kt4bbz

"AI layoffs" like Block's are a race to the bottom that have nothing to do with productivity. You won't be replaced by AI, nor a person using AI, because backfills aren't real.

Jack Dorsey is copying the 2023 leadership framework:
- overhire
- lay everyone off
- pretend it's because of AI productivity gains
- stock go up

The layoff victims were some of the most productive, AI-pilled people in the company, but it didn't save them, because that's not what it was about.

I am a slow synthesizer so I write better than I talk, unless I am public speaking, in which case I talk better than I write. In conversation, though, I kind of just stare at people and go "Y'know. So. Yeah. I dunno." I need, like, single-person-speaking training

Robert Smithson. A heap of Language. 1966 | MoMA Robert Smithson. A heap of Language. 1966. Pencil on graph paper. 6 1/2 × 22" (16.5 × 55.9 cm). Gift of Jan Christiaan Braun in honor of Agnes Gund and Marie-Josée Kravis in appreciation of their extr...

for Vilém Flusser, hens are also intelligent and astute readers (if not writers) in their ability to select signs from a heap--whether peas or corn kernels. both Flusser and hens would like Smithson's "Heap of Language" from 1966 www.moma.org/collection/w...

Source describes the failed Pentagon-Anthropic talks: through the end, the Pentagon wanted to use Anthropic's AI to analyze bulk data collected about Americans (Ross Andersen/The Atlantic)

Main Link | Techmeme Permalink

Now is when we need more dogs on the timeline most

They don’t need a tool to assess where to drop a bomb when they have one that confidently confirms any location they choose. They don’t need autonomous drones to drop those bombs, they just need to give an order. So the human is just a conduit for the machine, which is just a conduit for the human.

Which is to say that Claude is telling them what they want to hear, they know this, and use it precisely for that reason.

Remembering how the Bush admin leaked info to the NYT and then used that NYT reporting to publicly justify the invasion of Iraq: Intelligence laundering has a whole new meaning.

> Another of al-Jazari’s fantastical contraptions is of special interest to historians of science as it is regarded by many to be the first programmable “robot” in history.

www.nationalgeographic.com/history/hist...

First Film to Depict a Robot Discovered in Michigan Created in 1897, it's over two decades older than the term "robot."

First Film to Depict a Robot Discovered in Michigan

How Teens Use and View AI Just over half of U.S. teens say they've used chatbots for help with schoolwork, and 12% say they’ve gotten emotional support from these tools. Teens tend to view AI's future impact on their lives mor...

57% of American teens use AI to find information; 54% use it for help with schoolwork. In December, 30% reported using it daily. I would be curious to hear what actions folks propose we do to handle this. www.pewresearch.org/internet/202...

I'm a cognitive scientist with an interest in epistemic vigilance, and this essay that's been going around gave me pause.

I don't think it's straightforward to apply the concept of epistemic vigilance to interactions with LLMs, as this essay does.

🧵/

sbgeoaiphd.github.io/rotating_the...

“Radical Left AI Company”

This is an essential read.

Stochastic Flocks and the Critical Problem of 'Useful' AI Acknowledging that AI systems are advancing does not buy into hype, it sharpens the precision of critical thinking about their impacts, says Eryk Salvaggio.

You might not agree, but there’s a growing consensus, even among reluctant users, that LLMs *feel different.* I don’t think it undermines critique to ask about the stakes of those feelings — and of this new goal of “agential AI.” 🧵

So I think those are tokenization errors, ie, words and numbers are there for our sake, the machine just has token IDs that usually include more than one letter. So you can have a bunch of words and code because it’s a different level of granularity than counting or spelling!

The Illusion of AGI, or What Language Models Can Do Without Thought It is not simple stubbornness that LLMs are not “intelligent,” much less a form of “general” intelligence, writes Eryk Salvaggio.

one simplified but interesting thing that happened in class after looking a quote from @eryk.bsky.social 's recent piece was to ask students how much they thought their "knowledge" came from textbooks vs. every other experience they've had: the ratio split. Most agreed on around 75% exp / 25% text

This App Warns You if Someone Is Wearing Smart Glasses Nearby The creator of Nearby Glasses made the app after reading 404 Media's coverage of how people are using Meta's Ray-Bans smartglasses to film people without their knowledge or consent. “I consider it to ...

NEW: A hobbyist has created Nearby Glasses, an app that warns you if someone close by is wearing smart glasses. 404 Media spoke to the creator who said he was inspired by our coverage that uncovers how men are wearing Meta's Ray-Bans to covertly film massage parlor workers.

Stochastic Flocks and the Critical Problem of 'Useful' AI Acknowledging that AI systems are advancing does not buy into hype, it sharpens the precision of critical thinking about their impacts, says Eryk Salvaggio.

The promise of instant, personalized code has human effects. Not every problem is a coding problem but with agents, everything starts to look like one: “The blurry edge cases stop being the soft fibers of a social fabric. They become a technical nuisance.” 🧵

my hot take is that LLMs are steered into coding precisely because their failures are a lot harder to spot there

In so far as one denies what is, one is possessed by what is not, the compulsions, the fantasies, the terrors that flock to fill the void.

Stochastic Flocks and the Critical Problem of 'Useful' AI Acknowledging that AI systems are advancing does not buy into hype, it sharpens the precision of critical thinking about their impacts, says Eryk Salvaggio.

That bot that deleted emails of a Facebook “AI alignment specialist” was a vibe-coded piece of viral slopware. LLMs are being steered into coding because failures are presumed to be easier to spot. But sycophancy, “hallucinations” and projection of mind happen with code, even when it “works.” ⬇️

Yes 100%. This is why I think approaching design critically matters — this is engineered! That’s also why I link to Jenny Davis’ excellent work on affordances in the final paragraph.

Over the last few weeks AI discourse hand wringing about how we need to find a better way to substantively talk about the issues without immediately shutting down when acknowledging usefulness. This is the exact type of article we need to have that more nuanced grounded discussion.

AI is McDonald’s. That Will Be Good, Bad, and Catalytic for Civil Rights Advocacy. The current trajectory of AI may have similar effects on how people engage with computers as fast food has on how we engage with eating, writes David Brody.

AI is like McDonald’s. It’s a mass production tool transforming what we can do, but exploits labor, harms environments, and produces low quality goods. Civil rights advocates need to both fight the harms and shape the upside. We need to make AI better than McDonald’s, writes David Brody.

« Designing software requires a soft touch in deciding what values and priorities it encodes and how it distorts, discards, or misrepresents the data it processes. Tracing this requires technical literacy, but also judgment about how code complements or detracts from the problems it tackles.. »

Stochastic Flocks and the Critical Problem of 'Useful' AI Acknowledging that AI systems are advancing does not buy into hype, it sharpens the precision of critical thinking about their impacts, says Eryk Salvaggio.

To acknowledge the ways AI systems are changing and advancing does not buy into hype, it sharpens the precision of critical thinking about their impacts. To make sense of what AI does to people, you also need to understand what it does for them, writes Eryk Salvaggio. buff.ly/u3jbi3U

What remains urgently in dispute are the boundaries of utility: what usefulness means, for whom, and under what conditions? At what cost and from whom are benefits derived, and how are benefits and risks distributed? What decisions are quietly removed from public deliberation and handed to automated systems controlled by corporations, governments, and other institutions? That people are using language models doesn’t make criticism of them irrelevant. It makes it urgent.

Loved this article and especially this excerpt. Not enough people are considering that maybe utilization isn't divorced from design and purpose. Who benefits from the use of these products is just as important as who is harmed, and this problematic in part because it's rarely just the consumer.

Thank you!