zogwarg

My conspiracy theory is that he isn't clueless, and that his blogposts are meant to be read by whoever is his boss. In the case of using LLMs for automatic malware and anti-malware.

"Oh you want me to use LLMs for our cybersecurity, look how easy it is to write malware (as long as one executes anything they download, and have too many default permissions on a device) using LLMs, and how hard it is to do countermeasures, it took me over 42 (a hint?) tries and I still failed! Maybe it's better to use normal sandboxing, hardening and ACL practices, in the meantime to protect ourselves from this new threat, how convenient it's the same approach we've always taken"

Both. Humans are fundamentally a social animal, Rousseau's "State of nature" doesn't really exist.

Both society and humans are also the cure though:

• All individuals have the ability to discern and to choose good
• Society can teach what is good, and our tendencies to watch out for, and for the most part it also does this.

I don't believe the flaw can be eliminated, nor that the attempt would be ethical. Perfect is the enemy of good, you should teach people as best you can, but in the end still let them choose, anything else is thought-stopping cultish totalitarianism.

I like the quote from Terry Pratchett, (Granny Weatherwax)

And sin, young man, is when you treat people as things. Including yourself. That’s what sin is.

I think the worst parts of society, and innate "laziness" leads people to treat others (or yourself) as things, but that it's also innate to "know" not to treat others (or yourself) as things.

I don't believe the flaw is hopeless, even if it stays with us forever (at the individual and societal level).

The article almost looks like satire.

If all script kiddies waste their time trying to use generative AI to produce barely functional malware, we might be marginally safer for a while ^^. Or maybe this is the beginning of an entirely new malware ecology, clueless development using LLMs falling prey to clueless malware using LLMs.

Not every rationalist I've met has been nice or smart ^^.

I think it's hard to grow up in our society, without harboring a kernel of fascism in our hearts, it's easy to fall into the constantly sold "everything would work better if we just put the right people in charge". With varying definitions of who the "right people" are:

• Racism
• Eugenics
• Benevolent AI
• Fellow tribe,
• The enlightened who can read "the will of the people" or who are able to "carve reality at the joints"
• Some brands of "sovereign citizen" or corporate libertarianism (I'm the best person in charge of me!).
• The positivist invokers of ScientificProgress™

Do they deserve better? Absolutely, but you can't remove their agency, they ultimately chose this. The world is messy and broken, it's fine not to make too much peace with that, but you have to ponder your ends and your means more thoughtfully than a lot of EAs/Rationalists do. Falling prey to magical thinking is a choice, and/or a bias you can overcome (Which I find extremely ironic given the bias correction advertising in Rationalists spheres)

From reading the paper I'm not sure which is more egregious, the frameworks that pass code and/or use exec directly without checking, or the ones that rely on the LLM to do the checking (based on the fact that some of the CVEs require LLM prompt jailbreaking)

If you wanted to be exceedingly charitable, you could try and make the maintainers of said framework claim that "of course none of this should be used with unsanitized inputs open to the public, it's merely a productivity boost tool that you would run on your own machine, don't worry about possible prompts being evaluated by our agent from top bing results, don't use this for anything REAL."

The 100% mathematical PROVABLY_CORRECT proof of existence of the supernatural is at least funny.

It fails to prove dualism, which it then calls the supernatural for no adequately explained reason:

There is nothing new under the sun. Nothing a 3-lb-brain hominid does is impressive. Everyone dies and leaves behind nothing. If no God exists, all is infinitely meaningless. Fortunately, we can prove with mathematical certainty that the supernatural exists:

Would a 5-lb-brain hominid bring new things under the sun ? How about a 15-ton-brain corvid ? How about an acausal robot god wrought from all the ditherings found across the net ? If it is still so why are you so concerned with phrenology ?

1. You cannot be deceived that you are conscious.

So far so good, not too contentious, you need consciousness to be deceived, though I will note that it doesn't prove consciousness, only use definitions tautologically.

2. Consciousness, in itself, contains only that which you aware of.

No ? Not necessarily, that's overly egocentric. What about the Id ? What about collective consciousness ?

3. Consciousness is composed of perceptions and a perceiver.

A bit contentious, and not a very rigorous definition.

4. Perceptions are not composed of material things. Red is not a spectrum of light, nor a retinal activation, nor an optical nerve signal, nor a biochemical process in your brain: it is only the experience the perceiver calls “red”.

Qualia != Perceptions, but this is not the worst sin in this "proof".

5. The perceiver is not composed of material things. Neither quarks, nor atoms, nor molecules, nor cells, nor organs of the brain, nor the brain > itself experiences red. Associated processes happen, but only the perceiver experience red. To say that a material object “perceives” anything is a category error.

Does a perceiver without a body even exist ? I'm not really a monist myself, but this is clearly a leap.

6. Therefore, your consciousness undeniably exists, but it is not material.

Again does it exist untethered from the material ?

7. That which exists, yet is not material, is supernatural.

Hum no ? At best preternatural, and even then if you think the natural world follows Dualism, then the spiritual is still natural. I mean yes this arguing about definitions, but by god is this silly.

8. The supernatural exists.

QED.

Let's not forget the:

Ah! PotemkinTurd-4.0 is getting worse! Like it's starting to make all the same mistakes that PotemkinTurd-3.0 used to make! Honestly Poirot-2 is just as good now.

Cue to an answer from PotemCorp:

We haven't changes anything since the release of 4.0, but thanks we'll look into possible causes.

Like yes those a big Spaghetti monsters of RHLF and sad attempts at content filtering and/or removals of liability from PotemCorp, but isn't a much more rational explanation that the product was never that good to begin with, fundamentally random, and that sometimes the shit sticks and sometimes it doesn't?

And those shadows are just as sentient as we are, even if they don't depict the world, they convey a perception of a hypothetical world in which they are accurate!

Trying to grapple with the meaning consciousness through input/output is so close to being philosophical zombies type interesting, and yet so far and vacuous in what he actually says, that could apply to dice picking which color the sky is today. Also pretty hilarious that we would choose being WRONG, as a baseline (because LLM's are so bad) for outrospection, instead using the more natural cooperative nature of language. (Which machines fail at, which is maybe also why)

You can take solace in the fact that mxcl hasn't contributed to brew since 2012 (I guess the world ended) at least in terms of commits.

EDIT: Even if they are better at PR the mere fact that they would be onboard with tea and AI generated logos/descriptions foisted on projects that didn't ask for them, and acting confused when people are justifiably angry, shows a disturbing lack of care and consideration. (Paired with I take it incorrect installation scripts even)

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?

It remains infuriating how many agencies/governments simply don’t have reliable online public feeds, with information.

Especially annoying when the only place with info is twitter and/or Facebook.

(That doesn’t fully solve the community contributions, which should also be useful, but at minimum any info from official accounts SHOULD be posted elsewhere)

It's "fun" to see them fail to grasp that a journalist (or outsider) doesn't need to have read all their blogposts, and that "who talks to who" is basic journalism.

If only you read those glorious posts you would be enlightened, and if you somehow still disagree then you are either a liar, an NPC, or have not read them carefully enough, which I can prove by using shibboleths on our communities accepted doctrine.

It always boggles the mind when people fail to grasps others as being real.