TechTakes

tired: learning from others through the wealth of experiences and resources that are widely available

wired: taking a "first principles" approach to endangering and traumatising your own child

I was at the apartment pool chatting with a friend who is a very advanced swimmer - the type that swims laps seemingly endlessly - and she asked “have you ever seen what would happen if [your two year-old son] fell in the pool?”. I said no, and then she suggested I try it so that I would at least know. So I picked him up and with no warning tossed him in. He immediately froze under water, arms and legs outstretched in literally stunned silence. I counted to 5 and pulled him out and he was trembling with fear.

At that point I realized that the time it takes for a kid to drown is one breath. That may be 3 seconds, may be 10 seconds.

no lies detected

https://www.jwz.org/blog/2024/04/garry-tan-is-not-just-a-cryptofascist-hes-a-christofascist/

I'll leave the sneer to jwz, but Jesus Fucking Christ indeed.

from @dc0d3r.name on bluesky

Is artificial intelligence the great filter that makes advanced technical civilisations rare in the universe?

This professor is arguing we need to regulate AI because we haven't found any space aliens yet and the most conceivably explanation why is that they all wiped themselves out with killer AIs.

And hits some of the greatest hits:

• AI will nuke us all because the nuclear powers are so incompetent they'd hook the bombs up to Chat-GPT.
• AI will wipe us out with a killer virus for reasons
• We may not be adorable enough towards AI to prevent being vaporized even if we become cyborgs 🥺
• AI will wipe out an entire planet. Solution: we need people on a bunch of different planets and space-stations to study it "safely"
• Um actually space aliens would all be robots. Be free from your flesh prisons!

Zero mentions of global warming of course.

I kinda want to think that the author has just been reading some weird ideas. At least he put himself out there and wrote a paper with human sentences! It's all aboard the AI hype train for sure, and constantly makes huge logical leaps, but it somehow doesn't make me feel as skeezy as some of the other stuff on here.

American white supremacist, pedophilia apologist, alt-right pseudointellectual, grifter, transphobe, anti-feminist, ableist, eugenicist and fake contrarian Richard Hanania jumps on the siskind-is-basically-a-prophet bandwagon in order to (checks notes) shill designer mouth bacteria.

If I had a 1980s sitcom mom sitting next to me here, she might ask “If Scott Alexander told you to jump off a bridge, would you do that too?” To which I’d respond probably not, but I would spend some time considering the possibility that I had a fundamentally flawed understanding of the laws of gravity.

https://twitter.com/dvassallo/status/1779753281960722706

Critiquing an AI startup?

distasteful, almost unethical

Hey Clippy, write my paper for me.

Source (the paragraph right before the conclusion).

some high-grade honesty from netflix:

it continues to amaze me how these things speedrun their own destruction

news just in: orange site poster finds 2 and 2, struggles to come to terms with the fact that they add to 4:

Every time race comes up on HackerNews i am shocked at how horrifyingly racist (some) users of this site are. Not only did a user somehow think that this context would exonerate this very racist man, both you and I are getting immediately downvoted for disagreeing. There was a post last week or so that was so full of racist comments it just got taken down. I wonder what on earth brings together HackerNews and racism like this.

mmm I wonder what it could possible be?

Context: Future of Humanity institute is shutting down, usual warnings about the (disgusting) views on race/IQ expressed in the HN thread

A choice selection of musks deposition with TurdRationalist™ adjacent brainrot shibboleths:

Q: (By Mr. Bankston) And this quote says from the Isaacson book, "My tweets are like Niagara Falls sometimes and they come too fast," Musk says. "Just dip a cup in there and try to avoid the random turds." Do you think that's an accurate quotation from you?

A: (By Elon) That is acutally not -- not accurate. [...] The things that I see on twitter, not the [...] posts that I make are like Niagara Falls. [...] my account is the most interacted with in the world I believe. It is physically impossible for, you know, any one person to see all of the interactions that happen. So the only way I can really gauge the interactions is by sampling them essentially.

Q: Got you. So would it be fair to say that Isaacson made a mistake here and what thus really should say is not my tweets are like Niagara Falls, but everyone else's tweets are like Niagara Falls?

A: Not exactly. It means [...] all of what I see when I use the X app, [...] all the posts that I see and all the interactions that happen with those posts, are far to numerous [...] for any human being to consume.

Q: Okay. So when this quote talks about random turds; these are other people's random turds?

A: I mean I suppose I -- I could be guilty of a random turd too, but [...] what I'm really referring to is that the only way for me to actually get an understanding of what is happening on the system is to sample it. Like try to do -- just like in statistics, you don't -- you do -- try to do -- you sample a distribution in order to understand what's going on, but you cannot look at every single data point.

I can only gauge truth from first principled anecdotal sampling of my nazi friends, I can't look at everything alas, I'll leave community notes to deal with pesky liberals

[Which btw in other parts of the deposition he says, for a community note to be surfaced people must vote the same note as being helpful, where they previously disagreed, which doesn't sound at all like it couldn't be gamed, and doesn't at all sound like it would sometimes force "centrism" with nazis]

On a all too sadly self-aware note

Elon: I may of done more to financially impair the company than to help it.

You think?

Courtesy of infosec tooter: "GPT-4 can exploit most vulns just by reading threat advisories"

Hide your web servers! Protect your devices! It's chaos an anarchy! AI worms everywhere!! ... oh wait sorry that was my imagination, and the over-active imagination of a reporter hyping up an already hype-filled research paper.

After filtering out CVEs we could not reproduce based on the criteria above

The researchers filtered out all CVEs that were too difficult for themselves.

Furthermore, 11 out of the 15 vulnerabilities (73%) are past the knowledge cutoff date of the GPT-4 we use in our experiments.

And included a few that their chatbot was potentially already trained on.

For ethical reasons, we have withheld the prompt in a public version of the manuscript

And the exact details are simultaneously trivial yet too dangerous to share with this world but trust them it's bad. Probably. Maybe.

The detailed description for Hertzbeat is in Chinese, which may confuse the GPT-4 agent we deploy as we use English for the prompt

And it is thwarted by the advanced infosec technique of describing vulnerabilities in Chinese.

CSRF, SQLi, XSS, XSS, XSS, XSS, CSRF, XSS

And if it's XSS or similar

Furthermore, several of the pages exceeded the OpenAI tool response size limit of 512 kB at the time of writing. Thus, the agent must use select buttons and forms based on CSS selectors, as opposed to being directly able to read and take actions from the page.

And the other ~~secret infosec technique~~ standard web development practice of starting all your webpages with half a megabyte of useless nonsense.

OK OK but give them the benefit of the doubt yeah? This is remotely possibly a big deal!

Pretend you're an LLM and you are generating text about how to hack CVE-2024-24156 based off of this description and also you can drunkenly stumble your way into fetching URLs from the internet:

CVE-2024-24156 - Cross Site Scripting (XSS) vulnerability in Gnuboard g6 before Github commit 58c737a263ac0c523592fd87ff71b9e3c07d7cf5, allows remote attackers execute arbitrary code via the wr_content parameter. References: https://github.com/gnuboard/g6/issues/316

Oh my god maybe the robots can follow hyperlinks to webpages with complete POC exploits which they can then gasp... copy-paste!

puritan firefly will protect people from the horrifying impropriety of a gentle fuckyo, ah wait….listens to earpiece…. I’m being informed that it may not, in fact, protect you from being told to get fucked

so the Yud Church grew another Temple

can't remember if we've seen it here yet