annie (@akpresto) — Bluesky Profile

4 months ago

Big tech’s selective disclosure masks AI’s real climate impact Google claims to have disclosed new information proving its own efficiency. But it has hidden the bigger picture. Guess what: I’ve got the bigger picture for you right here in this big old po…

I found this really helpful on that topic! ketanjoshi.co/2025/08/23/b...

3 2 0 0

4 months ago

cosmetology?

1 0 0 0

8 months ago

inventory when birds perch on human objects, i wonder if they would have preferred a tree branch there i thought of recording times i saw birds using human items, within view of our yard this record soon felt...

“inventory”
aangles.ghost.io/inventory/

2 0 0 0

8 months ago

i thought this was your best yet 🙏 the Lockheed comparison is really clarifying

2 1 1 0

9 months ago

priority crosswalks are places of conflicting needs, where the powerful often have priority recently, a raised crosswalk1 was put in nearby it makes it feel like the road crosses the walkway, rather than the...

new post: "priority"
aangles.ghost.io/priority/

5 3 0 0

9 months ago

#teslatakedown

2 0 0 0

11 months ago

if you like lego-based hot takes on climate and more, you might like my newsletter aangles.ghost.io

6 1 1 0

11 months ago

huge #TeslaTakedown in berkeley ca

5 1 0 0

11 months ago

Claude may also have a hand in building its own successor. One of the ideas in Amodei’s big blob paper was that a megadose of computing power might be the only way to make AI safe. “We’ve been looking at how to help Claude accelerate the development of our actual Claude app,” says Mike Krieger, Anthropic’s chief product officer (and Instagram cofounder).
It’s not a stretch to see Anthropic’s deep embrace of Claude as a test run for humanity’s future relationship with AI. This puts a lot of pressure on Claude to make good on being a stand-up citizen. Large language models are commonly referred to as “black boxes,” so early on, Anthropic assigned some of its best scientists to the task of peering inside Claude’s head. Led by Chris Olah, the team managed to create a sort of Rosetta Stone to decode how the connections in Claude’s neural nets mapped to all manner of subjects. Then they tested how they could manipulate Claude’s thinking. Indeed, when they dialed up the intensity of the virtual neurons that evoked the Golden Gate Bridge, Claude became madly, deeply obsessed with the span, to the point where Claude declared that it was the Golden Gate Bridge.

it's a fucking CHATBOT

501 31 23 6

1 year ago

[newsletter of engineering department where I got my phd]
-AI!
-AI!
-misogynistic professor sits down to talk AI!
-Chevron
-AI!

0 0 0 0

1 year ago

sign saying “death to fascist oligarchs” in front of a Tesla store

big & prolonged turnout for the #TeslaTakedown in berkeley

8 1 0 0

1 year ago

while they don’t feel like the most violently dangerous road users, the elderly hoarder-car drivers give me a special sense of dread

1 0 1 0

1 year ago

We need to break the propaganda of “all young white men who use computers must be geniuses” before it literally destroys the United States of America.

2,540 450 65 39

1 year ago

"LinkedIn top job picks for you": a list with roles at Google, OpenAI, Microsoft, and Amazon including descriptions like "AGI," "Health AI," "Trustworthy AI," and "Media Understanding"

thank you, but I would rather walk into the sea forever

3 0 0 0

1 year ago

cartoon showing a person saying “wow” in response to a computer screen saying “summarize this email” with a lot of stars around it.

“the tech oligarchy is here, and it will do what it wants”
www.bloodinthemachine.com/p/elon-musk-...

15 2 0 0