He specifies Claude 10, so it’s after the ??? in the checklist, and spicy autocomplete will be widely accepted to qualify for sentience by then I guess.

Again he gets one-shotted by using an LLM helper to do the most trivial stuff, so it must be just as good at everything else, right?

These AI tools mix several techniques to disenshittify users’ devices. One major component is automating the installation of existing alternative firmware for these devices that have been (possibly illegally) created by hobbyists and uploaded to various code repositories. Installing these alternative OSes is typically a gnarly task, requiring the user to do a lot of complex, fiddly, highly technical work to bypass their devices’ digital locks.

It it’s anything like any android device I’ve ever put custom stuff into, the gnarly and highly technical work alluded is just following a checklist.

Much of what a vibe-coded jailbreaking tool does is stand in for Canonical, but for hobbyist OSes for smart TVs and speakers and many other devices.

groan.

It kind of devolved to LLMs are so cool guys type stuff where he brags about his chatbot helping him with his linux power management by unearthing an undocumented environment variable, so when other people say it wrote a custom os for their device from scratch that must have been just as frictionless, right?

I don’t know what I’m supposed to get from this piece, it feels like he’s saying LLMs are about to solve enshittification as long as you’re willing to take the responsibility of e.g. using a hella unsafe os, so basically what’s left if repealing the laws that make tampering with your device this way broadly illegal.

There’s also a part in the denouement where he says a possible solution to improving slop code quality and safety standards is uploading the code to public repositories so similarly inclined people can engage with it, which, I don’t know, seems kind of familiar if you’ve been a programming adjacent for two seconds. But then you’d also be aware what the slopnami is already doing to the open source ecosystem.

The last slatestar was about AI successionism. As in, No no no, we’re definitely not like those freaks who think humanity should just cede primacy to the first AI that shows us a bit of leg. Anyway here’s how we should actually surrender to chatbots, cue rant about Kemal AI, how we could change the constitution so Claude can run for president, or have an AI legislator body were interested groups train their own AIs according to political preference and let them work it out between themselves, because politics and legislation are famously incredibly clear cut, no gray areas, no necessary compromises, no interpretability, no irrational actors, no incompatible agendas, everyone is pleased, make no mistakes.

He almost gives away the game when he says christians don’t feel like slaves when they say ‘thy will be done’ so letting Claude run society would feel fine.

The WSJ apparently fucked up describing how Anthropic is losing money and now the discourse is very messed up with a lot of people claiming they are breaking even or even actually making money, basically the AI booster ecosystem is having a huge HA HA GOT EM! The fiinancials were always better than the lying press made it seem! moment.

In a new draft prospectus viewed by the Journal, the AI giant disclosed an $80.7 billion net loss for the first half of 2026 and a long list of risk factors

According to my degree in finance from Very Online University the 80 bil net loss of the article’s summary ends up consisting of 72B of convertible notes, which I guess is a type of debt that goes away if the IPO goes well because it converts to equity. Also because it is predicated on equity its nominal value goes up as the projected valuation of the company goes up, so the 72B figure doesn’t even mean that much. Supposedly for Anthro to be on the hook for even a portion of that they would have to IPO at a lower price than expected when this type of debt was originally issued, which probably isn’t happening.

Decent monday morning take from robert evans:

Schrodinger’s double-think.

other ai pushers declaring the ai conscious

Hey now, segyges has had his own whinging multithousand word blogpost on model consciousness since early spring (tldr: currently not really, but it’s inevitable because I really want it to, so don’t be on the wrong side of history when it happens).

He’s seems to be angling for the role of TPOT scott alexander, except instead of gradually cultivating a consensus in favor of eugenics and being ruled by high IQ overmen his long term thing is establishing any perceived anti-AI sentiment as an unforgivable moral failing.

I could have gone with destroying excess capital to preserve the status quo, but this stage and with this administration in the US going for subtlety seems unnecessary .

Robin Hanson seems like the incredibly obvious primary suspect, you mean where he got the idea in the first place?

He is a close associate of Eliezer Yudkowsky, who formerly co-blogged with Hanson on Hanson’s blog, Overcoming Bias, before starting LessWrong.

RW also says 2013 was when he first published as papers all his prediction market stuff.

Programming is hard. So, instead of doing that, we vibe coded a new Epic Games Launcher that’s way faster and more responsive than the current one, and has a new look.

Just pure fellow vibecoders cringe.

I really liked the part about all countries being in debt, straight from the facebook comments school of economics.

It’s a catchy summary because he leans into the weird sex stuff instead of say the cosmism aspect, but the writer is a knob who doesn’t like EA/rats for basically the same reasons as Peter Thiel, so take him with a grain of salt.

He’s also likened anti-data center activism to jewish pogroms and believes that AI resentment can be summed up as “curing cancer is less important than my job satisfaction”.

What type of documentation are you looking for? The big picture is easily googleable and wikipediable: EY got a few of the super-rich excited enough about doomer drama to bankroll the founding of OpenAI, whose governing body was mostly EAs (until the Altman-Nadella counter-coup) while Anthropic were the people who bailed because they though OAI weren’t serious enough about alignment, and suddenly getting dicked in a rationalist compound group house has immense career implications in the burgeoning AI sector.

Also terms like AGI, ASI, AI alignment, existential risk et al started out as lesswrong shibboleths, and so-called alignment research is almost completely a rationalist circlejerk.

Pinning the EA/rats down as a cult is where it gets murkier because it’s mostly them telling on themselves in social media and blog post implications, like how they foster a culture of zero personal boundaries, ingroup overdependence and rigid hierarchies stratified around claimed IQ and proximity to the Great Work of not messing up the robot god incantation that ends the world.

You could do worse than the Behind the Bastards episode on the Zizians, mostly because unlike Leverage or MAPLE who kind of did their own thing but with rationalist funding and human capital, the Zizians can be fairly described as what happens if you actually take the more esoteric aspects of rationalism seriously.

edit: failed to spot the ninja edit so the response made no sense

Anthropic has a new angle for trying to get local LLMs banned, apparently they’re about to be able to use your temu android tv to take over your smart toaster:

Our experiments suggest that small AI models may be months away from being capable enough to self-replicate on commercial internet-of-things (IoT) devices, like smart ovens and off-brand Android TV boxes. A replicator like this would be challenging to shut down, since there’s no centralized point to shut off—though it may be slow (with a doubling time of days). Unlike non-AI malware, it could also figure out unfamiliar devices on its own. In our evaluations, a 9B model got inference running on 8 of the 9 hardware types we tested

edit: The TVs need to be off-brand to lack basic security features, and also there must be at least four of them to even start being inference-capable, and the actual setting up of a distributed inference network on android tvs is handwaved as “feasible”.

The thrilling conclusion of this blog post is called Signs of Life, because of course it is.

segyges already trying to launder the unhinged out of the our chatbot will be curing cancer any minute now new round of claims by framing widespread resentment against AI as “curing cancer is less important than my job satisfaction”.