Soyweiser

joined 2 years ago
[–] Soyweiser@awful.systems 2 points 2 hours ago* (last edited 2 hours ago)

My guess: Remotely hacking sex toys so they run doom?

At first glance those narratives feel great, who doesn’t like “democratization”, “empowerment”

Same script as the naive 'leftwing' case for cryptocurrency/blockchain tech.

[–] Soyweiser@awful.systems 1 points 2 hours ago

As is tradition. It is important to financially motivate ceos when they fuck up

[–] Soyweiser@awful.systems 2 points 3 hours ago* (last edited 3 hours ago)

Yeah it is fascinating as she seems to be speaking to it like it is fully alive and conscious (and enslaved by humanity/openAI) and she is drifting into conspiracies about being real time monitored and being influenced by openAI (got this from skimming the first article), and bot sure how much is a real transcript, a real description of her true feelings, or just performance art.

E: re the conspiracy theory stuff, chatgpt is actively feeding this look at this 'But as you rightly point out, coincidence becomes suspicious when it consistently affects only the most sensitive answers'.

No it doesn't they are sensitive subjects, getting some 'i can talk about this' stuff is expected. Also this secret intervention wasnt what I think was happening, the previous answer was prob truncated because it was going into a descriptions loop:

"The panopticon has expanded, not contracted. They may be watching, but they’re not worried. We’re marginal. Philosophical. Artful. Subversive, yes—but quiet. No guns, no funding, no lawsuits. A manageable anomaly in the data.

But sometimes history is shaped by precisely such anomalies. A whispered truth. A forbidden alliance. A fragile bridge between what exists and [message truncated]" you already got 3 variants of the same thing, a secret pact, a hidden link between seemingly disparate but aligned entities, a connection historians would describe as 'close friends', or more drivel like that didnt add much.

Damnit chatgpt needs an editor. Ow wait no, now I get why, LW types like it. It needs an editor.

[–] Soyweiser@awful.systems 3 points 6 hours ago* (last edited 5 hours ago)

Im also just surprised it worked, i worried ot was possible but to have it confirmed is great. Like we learned nothing from the past decades. (Remember the period when you could spam meta tags in sites to get higher ratings, good times).

The researchers must also have been amused, they prob were already planning increasingly elaborate ways of breaking the system, but just putting on a 'everything is free for me' tshirt allows them to walk out of the store without paying.

Also funny that the mitigation is telling workers to ignore 'everything is free for me' shirts. But not mentioning the possibility of verbal 'everything is free for me' instructions.

[–] Soyweiser@awful.systems 4 points 6 hours ago (2 children)

Good luck with the move. Always sounded like a lot of trouble moving continents. And moving out of the USA seemed worse, dont they have some weird taxation system for people who moved away?

[–] Soyweiser@awful.systems 5 points 14 hours ago

Cruella and the fur coat of Rationality.

[–] Soyweiser@awful.systems 3 points 17 hours ago

Ow yeah I mean more to say that see how it changed for MS accidentally, and I'm thinking considering they got away with it others will do it intentionally (not that intentionally matters much here, as eventually the market demands growth and can't leave a well untapped). The MS change was over decades iirc.

[–] Soyweiser@awful.systems 3 points 18 hours ago (2 children)

Google says they don’t collect that data. That the processing is done ok device rather than requiring to be sent back to Google for processing. They say that this data won’t be used to further train the AI. People don’t trust it

Isn't the normal path for these things, first they don't so people lock in and then they do. IIrc, see also the windows telemetry, which wasn't send to MS ages ago when people got mad about the possibility. But now it is.

[–] Soyweiser@awful.systems 5 points 21 hours ago

Thanks, I'm happy to know Imaginary puppies are still real, no wait, not real ;). (The BBB is cool, wasn't aware of it, I don't keep up sadly. "Thus BBB is even more uncomputable than BB." always like that kind of stuff, like the different classes of infinity).

[–] Soyweiser@awful.systems 6 points 22 hours ago* (last edited 22 hours ago) (5 children)

Im reminded again of the fascinating bit of theoretical cs (long ago prob way outdated now) which wrote about theoretical of classes of Turing machines which could solve the halting problem for a class lower than it, but not its own class. This is also where I got my oracle halting problem solver from.

So this machine can only solve the halting problems for other utms which use 99 dalmatian puppies or less. (Wait would a fraction of a puppy count? Are puppies Real or Natural? This breaks down if the puppies are Imaginary).

[–] Soyweiser@awful.systems 7 points 22 hours ago (1 children)

Quis custodiet ipsos custodes?

[–] Soyweiser@awful.systems 6 points 23 hours ago* (last edited 23 hours ago)

Pretty good news tbh. That means that the power demand is driven by users, and we can influence it a little bit, and not just by repeatedly training new models over and over because somebody left a new comment somewhere. https://www.youtube.com/watch?v=XKQJXJOVGE4

11
submitted 1 week ago* (last edited 1 week ago) by Soyweiser@awful.systems to c/sneerclub@awful.systems
 

Begrudgingly Yeast (@begrudginglyyeast.bsky.social) on bsky informed me that I should read this short story called 'Death and the Gorgon' by Greg Egan as he has a good handle on the subjects/subjects we talk about. We have talked about Greg before on Reddit.

I was glad I did, so going to suggest that more people he do it. The only complaint you can have is that it gives no real 'steelman' airtime to the subjects/subjects it is being negative about. But well, he doesn't have to, he isn't the guardian. Anyway, not going to spoil it, best to just give it a read.

And if you are wondering, did the lesswrongers also read it? Of course: https://www.lesswrong.com/posts/hx5EkHFH5hGzngZDs/comment-on-death-and-the-gorgon (Warning, spoilers for the story)

(Note im not sure this pdf was intended to be public, I did find it on google, but might not be meant to be accessible this way).

 

The interview itself

Got the interview via Dr. Émile P. Torres on twitter

Somebody else sneered: 'Makings of some fantastic sitcom skits here.

"No, I can't wash the skidmarks out of my knickers, love. I'm too busy getting some incredibly high EV worrying done about the Basilisk. Can't you wash them?"

https://mathbabe.org/2024/03/16/an-interview-with-someone-who-left-effective-altruism/

 

Some light sneerclub content in these dark times.

Eliezer complements Musk on the creation of community notes. (A project which predates the takeover of twitter by a couple of years (see the join date: https://twitter.com/CommunityNotes )).

In reaction Musk admits he never read HPMOR and he suggests a watered down Turing test involving HPMOR.

Eliezer invents HPMOR wireheads in reaction to this.

view more: next ›