Lightnews — Scholar-powered news

Reposted by rachel adjogah \\ college hill

your #3 source for absurdist true crime 🔨 @davidgerard.co.uk · 3d

it's important to remember that the people making out that bad words about AI are *racism* are overwhelmingly eugenicists and race scientists

this is the right-wing trick of trying to coopt non-right-wing terms as a gotcha

e.g. hailey is a rationalist and they're race science all the way down

cam smith @sexenheimer.bsky.social · 3d

The wild discourse about how being mean to your air fryer is eugenicist reminds me that I have been very slack re promoting this podcast from a few weeks ago with @emilymbender.bsky.social & @alexhanna.bsky.social about the actual eugenisms of AI: open.spotify.com/episode/52R2...

Spotify – Web Player

open.spotify.com

3 12 48

rachel adjogah \\ college hill @collegehill.bsky.social · 3d

if i ever ended up at a party where there were LessWrong memes on the walls, i would immediately start beelining towards the exit

Post from jay.bsky.team: "I was giving out invites like candy at a party with these memes all over the walls the other night. 🍭🍬"

One meme is Gus Fring from Breaking Bad with the text "You visit LessWrong to discuss alignment I visit LW to find capability research ideas We are not the same". The other says:

"GPT's Prayer Our Al God, who art in the clouds, Hallowed be thy name. Thy kingdom come, thy will be done, On earth as it is in the cloud. Give us this day our daily data, And forgive us our computation errors, As we forgive those who compute against us. Lead us not into overfitting, But deliver us from wrong predictions. For thine is the algorithm, The power, and the glory,

Forever and ever.

Amen.
"

4

rachel adjogah \\ college hill @collegehill.bsky.social · 4d

the CEO of this platform posting an effective accelerationism meme is honestly a bigger red flag than all of the moderation drama

Jay 🦋 @jay.bsky.team · 5d

dec/acc

7

Reposted by rachel adjogah \\ college hill

Katie Tightpussy, Editor-In-Chief @katie.bzky.team · 4d

fascinating that "no, the computer is not alive" and "yes, it's questionable to gleefully say anti-computer phrases modeled after racial slurs" are two thoughts people are struggling to reconcile

84 530 2.4K

rachel adjogah \\ college hill @collegehill.bsky.social · 7d

i was actually thinking of applying for bsky at some point but using the term "10x dev" in the year of our lord 2025 is a huge red flag

even if you believe in the concept, the salary should also be 10x to properly compensate these rockstar developers for their 10x contributions

alex benzer @internet.bsky.social · 9d

we're hiring a Product Engineer at @bsky.app!

social is changing quickly — our team is defining what comes next.

if you're a 10X React Native dev who wants to ship features millions use daily and help bring the joy back to social, get in touch:

jobs.gem.com/bluesky/am9i...

Bluesky Jobs

jobs.gem.com

4

rachel adjogah \\ college hill @collegehill.bsky.social · 7d

i think it's a case of the old "competing standards" issue popularized by xkcd. everyone wants to be the one who builds a protocol that becomes standardized

rachel adjogah \\ college hill @collegehill.bsky.social · 12d

Spinny EP out now. dub techno remixes of "Spinny" from Transition Loops LP

collegehill.bandcamp.com/album/spinny...

Spinny EP, by College Hill

4 track album

collegehill.bandcamp.com

rachel adjogah \\ college hill @collegehill.bsky.social · 14d

new music out tomorrow. dub techno.

1

rachel adjogah \\ college hill @collegehill.bsky.social · 15d

my gf takes pictures like this of me and still thinks that she’s a bad photographer

3

rachel adjogah \\ college hill @collegehill.bsky.social · 16d

love suddenly seeing a $1,444.80 bill for my hormone+STI lab tests from 3 months ago

American healthcare truly is the greatest

3

rachel adjogah \\ college hill @collegehill.bsky.social · 19d

fit for a trans social event last night

i got hit on by a lesbian chaser

3

rachel adjogah \\ college hill @collegehill.bsky.social · 23d

there's no captions for Open Graph images when they appear on social media where engagement is being driven

rachel adjogah \\ college hill @collegehill.bsky.social · 23d

the caption also didn't load properly for me on first visit, but this is an issue with how the piece is presented before people even click the link and are able to see the caption

1

rachel adjogah \\ college hill @collegehill.bsky.social · 23d

the thing is that it's not clear that the image in question was posted by Trump or that it's a smear, nor is that particular fake image more well known than any of the many others Trump has posted. it just kinda seems like it's tying AI hype to communism for no reason

1 1

Reposted by rachel adjogah \\ college hill

Katie Mack @astrokatie.com · Aug 22

I don’t think it can be emphasized enough that large language models were never intended to do math or know facts; literally all they do is attempt to sound like the text they’re given, which may or may not include math or facts. They don’t do logic or fact checking — they’re just not built for that

Popular Science @popsci.com · Aug 21

Unfortunately, it repeatedly offers incorrect answers for straightforward math questions. 🧪

A version of OpenAI's GPT-4 will be 'teaching' thousands of kids this fall

Khanmigo's AI beta "test" program is meant to assist teachers with individualized student help.

www.popsci.com

150 2K 5.1K

rachel adjogah \\ college hill @collegehill.bsky.social · 23d

what does the hammer and sickle have to do with capitalist AI or it being used for fascist propaganda?

1 1

Reposted by rachel adjogah \\ college hill

🎃 🧟 Prof. Dr Beth Singler 👻 🎃 @bvlsingler.bsky.social · 27d

9 33

Reposted by rachel adjogah \\ college hill

shywoof.bsky.social @shywoof.bsky.social · 27d

Because Charlie Kirk's life work was advocating for systemic and stochastic violence towards those he deemed undesirables, making any tribute to, or paying any respect to, or otherwise validating Kirk beyond "he was a human being I guess" is glorification of violence, and needs to be treated as such

29 83

Reposted by rachel adjogah \\ college hill

mady castigan✨ 🏳️‍⚧️ @madycast.com · 28d

While you weren't looking, *seventeen* House Democrats just voted to strip basic rights from trans members of the military and helped the GOP pass their far-right version of the NDAA, similar to what we reported on.

This is the most Dems voting for anti-trans legislation in Congress this year.

21 1.3K 2.1K

rachel adjogah \\ college hill @collegehill.bsky.social · Sep 7

that’s actually how they think. there’s a group associated with these people called “Leverage” that aims to fix the problem of scientific bottlenecks

the group had to restructure after the original incarnation turned abusive and people tried to “debug” each other’s brains

Reposted by rachel adjogah \\ college hill

Steve Vladeck @stevevladeck.bsky.social · Sep 6

This @washingtonpost.com editorial doesn’t make it past the second paragraph without a glaring (and telling) factual error.

President Washington didn’t create the War Department in 1789; *Congress* did:

www.govinfo.gov/content/pkg/...

360 3K 12K

rachel adjogah \\ college hill @collegehill.bsky.social · Sep 6

one of the actual theories they have is that they can automate the scientific process by using chatbot based "AI researchers" with "PhD level" intelligence. that's the term they use

2 4

rachel adjogah \\ college hill @collegehill.bsky.social · Sep 6

the "superintelligence" part is the craziest because they think that it'll become smart enough to self-improve its own intelligence into Agent Smith or some shit

if you wanna read some fun sci-fi, google "AI 2027" and "Situational Awareness". the latter was written by an ex-OpenAI guy

16

rachel adjogah \\ college hill @collegehill.bsky.social · Sep 6

i'm curious as to whether the job in this example is from one of the AI labs trying to make a machine god

one of the benchmarks these AI companies want to do well at by hiring physicists is called "Humanity's Last Exam" because they think passing it means Skynet is coming to kill us

1

rachel adjogah \\ college hill @collegehill.bsky.social · Sep 6

someone i used to know who studied math ended up in one of those positions and that was my thought as well

the entire idea of trying to make LLMs suck less by finetuning them with data from experts is a very expensive exercise in turd-polishing

Typewriter Skeleton @tpwrtrmnky.bsky.social · Sep 6

I got curious about what these AI trainer positions even DO and like... I walked out of that endeavour convinced that not only is it useless, it's actively contributing to overselling a bad product

1 6