TechTakes

Good news everyone, Dan has released his latest AI safety paper, we are one step closer to alignment. Let's take a look inside:

Wow, consistent set of values you say! Quite a strong claim. Let's take a peek at their rigorous, unbiased experimental set up:

... ok, this seems like you might be putting your finger on the scales to get a desired outcome. But I'm sure at least your numerical results are stro-

Even after all this shit, all you could eek out was a measly 60%? C'mon you gotta try harder than that to prove the utility maximizer demon exists. I would say our boi is falling to new levels of crankery to push his agenda, but he did release that bot last year that he said was capable of superhuman prediction, so this really just par for the course at this point.

The most discerning minds / critical thinkers predictably reeling in terror at another banger drop from Elon's AI safety toad.

*** terrifying personal note: I recently found out that Dan was my wife's roommate's roommate's roommate back in college. By the transitive property, I am Dan's roommate, which explains why he's living rent free in my head

Saltman has a new blogpost out he calls 'Three Observations' that I feel too tired to sneer properly but I'm sure will be featured in pivot-to-ai pretty soon.

Of note that he seems to admit chatbot abilities have plateaued for the current technological paradigm, by way of offering the "observation" that model intelligence is logarithmically dependent on the resources used to train and run it (i = log( r )) so it's officially diminishing returns from now on.

Second observation is that when a thing gets cheaper it's used more, i.e. they'll be pushing even harded to shove it into everything.

Third observation is that

The socioeconomic value of linearly increasing intelligence is super-exponential in nature. A consequence of this is that we see no reason for exponentially increasing investment to stop in the near future.

which is hilarious.

The rest of the blogpost appears to mostly be fanfiction about the efficiency of their agents that I didn't read too closely.

The singularity will be a dozen crappy models in a trench coat, and then finally we’ll have Magic Unified Intelligience™

In a hilarious turn of events that no one could have foreseen, Anthropic is having problems with people sending llm generated job applications, and is asking potential candidates to please not use ai.

While we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process. We want to understand your personal interest in Anthropic without mediation through an AI system, and we also want to evaluate your non-AI-assisted communication skills. Please indicate 'Yes' if you have read and agree.

https://www.404media.co/anthropic-claude-job-application-ai-assistants/

In other news, all hell's broken loose at BlueSky: https://bsky.app/profile/culturecrave.co/post/3lhv35la2pk2h

The "AI company" in question is a nonprofit that focuses on open-source safety tools which recently launched at the Paris AI Action Summit, but that was enough to cause things to go nuclear, especially given people initially flocked to BSky to get away from AI.

Thinking I should make this into a full post.

An entertaining bit of pushback against the various bathroom bills being pushed at the moment. Bonus points for linking it with ai training. I feel like this is an idea that’s very adaptable…

https://mefi.social/@MissConstrue/113983951020093710

Signs which have been adhered to bathroom stall interiors at the Dallas Fort Worth airport.

SECURITY NOTICE Electronic Genital Verification (EGV) Your genitalia may be photographed electronically during your use of this facility as part of the Electronic Genital Verification (EGV) pilot program at the direction of the Office of the Lieutenant Governor. In the future, EGV will help keep Texans safe while protecting your privacy by screening for potentially improper restroom access using machine vision and Artificial Intelligence (Al) in lieu of traditional genital inspections. At this time, images collected will be used solely for model training purposes and will not be used for law enforcement or shared with other entities except as pursuant to a subpoena, court order or as otherwise compelled by legal process. Your participation in this program is voluntary. You have the right to request removal of your data by calling the EGV program office at (512) 463-0001 during normal operating hours (Mon-Fri 8AM-5PM). STE OP CRATMENT OA Pusi DFW DALLAS FORT WORTH INTERNATIONAL AIRPORT

The contact number appears to be for Dan Patrick, the lt. governor of Texas.

Niko Matsakis wrote a post about how the fucking rust compiler toolchain should include an LLM to explain error messages because teaching the semantics of a language is too hard and that pissed me off so much that instead of linking that piece of shit directly I’m posting this excellent sneer from Anatol Ulrich that’s also much shorter than Niko’s extended attempt to beg for a promotion at Amazon

e: mastodon thread

US government tech hellscape roundup part the third (ugh):

1. Elon Musk jokes(?) that the government doesn't use SQL ??? (source, note that his tweet has an ableist slur). I don't even know what to think about this. Is it supposed to be funny or something? Does he actually believe it?

2. Article: Elon Musk’s A.I.-Fuelled War on Human Agency -- People here probably already knew all this; but one of the ways the admin thinks they can fire everyone is by replacing people with AI / automating everything. Some of the social media responses from federal workers are pretty great:

   Really excited to see AI put on some waders and unclog a beaver dam from a water structure for me.

   If I've learned anything from all this it's about how ~~unfathomably based~~ cool a lot of federal workers are.

3. The less fascist / cowed parts of the infosec industry are currently raising the alarm about how insecure this all is. A representative social media post from Gossi The Dog

   I definitely recommend posting about what is happening in the US on LinkedIn as you will quickly learn many of the largest security vendors are staffed by people who have no interest in protecting people, while posting with their employers names.

4. Some federal workers have been fired via emails calling them [EmployeeFirstName].

Edit:

5. ~~Elon Musk~~ The US State Department plans to buy $400m worth of armored Cybertrucks from Elon Musk (nytimes) (Edit: may have been ordered under Biden's administration)

6. doge.gov has been updated. Mostly just with more useless baby's first website materials; but they promise a "comprehensive, government-wide org chart" and are hiring "software engineers, InfoSec engineers, and other technology professionals".

   Aside: I already found ~~two~~ three minor website bugs despite not really looking for them and the website being tiny. But that can't be right... they're IT professionals while I'm DEI.

7. Find replace is so hard :( and that's why the government writes about "gay and rights" to avoid saying the... the... the forbidden t-word of which I dare not speak

8. So about how I said doge.gov gives baby's first website vibes; it's database was left world writable lol

9. doge.gov shares classified information

10. Classic Musk "humor": a "tech support" T-shirt to allude to all of this. The dude really likes custom T-shirts (which to be fair custom t-shirts can be awesome when they're less bad)

Holy smokes Jeeps will reportedly show ads while you are freaking driving:

Imagine pulling up to a red light, checking your GPS for directions, and suddenly, the entire screen is hijacked by an ad. That’s the reality for some Stellantis owners. Instead of seamless functionality, drivers are now forced to manually close out of ads just to access basic vehicle functions.

One Jeep 4xe owner recently shared their frustration on an online forum, detailing how these pop-ups disrupt the driving experience. Stellantis, responding through their “JeepCares” representative, confirmed that these ads are part of the contractual agreement with SiriusXM and suggested that users simply tap the “X” to dismiss them.

"Listen guys, if you don't want me stabbing you you simply have to ask nicely every time, and also I'm trying real hard to reduce the rate of stabbing incidents so in a way I'm the victim here."

Reading around it sounds like modern cars can be user-hostile in general, and this might not be new; so I'm sure glad I have one from the ancient times of 2012. It has a tiny unobtrusive screen which does nothing but show my music, the odometer, the backup camera, any warnings, and the Hatsune Miku wallpaper I loaded into it.

the scott alexander fanclub on reddit is especially unhinged lately. Some choice picks, beginning gently with a sneer:

Anyone who says there is a 0% chance of a fried egg being sentient is overconfident. Nobody knows what causes consciousness. We have no way of detecting it & we can barely agree on a definition. So we should be less than 100% certain about anything to do with consciousness and fried eggs.

moving on,

For me, I currently believe that the way forward is gathering my Dunbar community of ~150 values-aligned people and dropping out together to live in intentional community, generally off-the-grid, interacting with technology only as a tool (no news/algorithms/push notifications), close enough to a major city to visit friends but far enough away to isolate ourselves when we want to.
We then push forward into an unknown world, supporting each other, doing enough trade and consulting with the outside world to ensure a positive trade surplus for the collective and distributing the profits among us to sustain the community.

Yes, that is a response to AI despair. In a "non-doomer thread". Because of course.

but what takes the cake is the regularly scheduled IQ thread, a beautiful dumpster fire which burns especially bright on this day. Yet one specimen soars above the din:

The average should recognize and accept their inferiority and make peace with the fact that they will live and die a mediocre life. They should stop trying to interfere in the affairs of their betters and content themselves with their own garden, which can provide enough happiness for them until the end of their life if properly cultivated.

Further study of the user's excreta clarifies that they are:

• sincere
• definitely not racist

warning: HEAVY PSYCHIC DAMAGE and I really hope the spoiler block functions correctly on your end

(archives: one two three)

Well as they promised Google Maps has finally fallen. It now shows "Gulf of America" and nothing else to US users. I suspect someone outside the US will be shown both the real name and Gulf of America. Denali is still labeled as Denali... for now.

Disorganize the world's information and make it universally inaccessible and stupid.

Idea for another megathread: go to linkedin and post the first thing you see (that provokes a reaction). Here’s mine:

FWIW I checked a few comment threads and guy is playing this off as lighthearted/a joke, but folks here know better than that.