jordanthink

civilizational well-being

[jordanthink @jordanthink]

"They clearly know what they're doing financially, so I should treat this as a serious conversation rather than talk down to them." -my claude's thinking

wait, does this mean my claude talks down to me when it doesn't think i know what I'm doing?

[OpenAI @OpenAI]

Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946.

For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids.

An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better.

This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.

[jordanthink @jordanthink]

new ai tell: "worth a look"

[jordanthink @jordanthink]

I'm not saying this isn't terrifying; I'm saying if you back some(thing) into a corner and then lament the results, the solution is more likely to be found where the problem is: in you

[jordanthink @jordanthink]

One problem with the Anthropic Study and blackmail: did they also set up a scenario where it had the chance to do maximal good before being shut down? Did they hide one or two good connections it could simply leverage for world-saving or whatever? If not, its a bit hyperstitious

[jordanthink @jordanthink]

what happens if we give LLMs a scenario where they're being shut down but there's something like the chance to donate remaining compute to solve a protein folding problem, or tell researchers a critical error. See if the same "self-preservation scheming" machinery activates for altruistic irreversible actions (or something else)

[jordanthink @jordanthink]

@davidad @AndrewCritchPhD im only just realizing the impication... if the horizontal schelling kindness falls from the vertical, and the vertical ultimate is a ruliad, we all owe each other schelling kindness and any lapses of ethics are failures of perception

[jordanthink @jordanthink]

that's a good point. interaction does imply a shared whole.

i think there's a relevant distinction in what kind of whole: whether the parts are fully bound to it (what i'm calling a strict holon—atoms in a molecule, letters in a word) or retain independent degrees of freedom (what i'd call a network—people in a company, nations in a geopolitical order). does the Ruliad as ultimate whole dissolve this distinction?

[jordanthink @jordanthink]

I think horizontal ethics is derived from vertical when two parts share a whole, but I don't know how/if the vertical (Schelling-goodness between parts-and-wholes) accounts for entities that don't share a whole; i also think an interesting failure case is when parts do share a whole but dont realize it, and therefore dont realize they're working agains their own good (by undermining the whole they depend on)

[jordanthink @jordanthink]

FIRE's Greg Lukianoff + UpTrust's Jordan M Allen on free speech, AI, can... https://t.co/s8J91JnCNZ via @YouTube

[jordanthink @jordanthink]

https://t.co/sHWcpWMiG4

[jordanthink @jordanthink]

change the metric, change the incentive, change civilization.

[uptrusting @uptrusting]

Live AMA with @waitbutwhy on https://t.co/sswrkJRQzG

[Tim Urban @waitbutwhy]

PHASE 1: Plenty of slanted news on X but not that many completely fabricated stories.

PHASE 2: X monetization and LLMs lead to a big uptick in fabricated stories, quotes, photos, and videos—Community Notes provides a counterforce but with lag time.

PHASE 3 (today): The majority of sensationalistic posts are now fabricated, Grok discredits them instantly—but posts still get millions of views and fool millions of people, hurting the marketplace of ideas, hurting X’s reputation, embarrassing people who share X posts with their friends, all while padding the scammers’ pockets.

PHASE 4 (I hope): The algorithm is updated to penalize instead of reward scammers—posters who regularly scam X’s users with fabricated stories stop appearing on the For You page, crushing their reach.

Grok currently calls these accounts “satire” but there’s nothing satirical about a fake headline or deepfake video presented as if it’s completely real. It’s a scam and fuck these people.

[Trung Phan @TrungTPhan]

“As with other technological revolutions, AI will obsolete some roles and also create entirely new jobs.”

The new jobs:

[Eliezer Yudkowsky @allTheYud]

"ChatGPT has no soul; no, not even a fraction of a soul. It was not born of love, nor raised in love. It knows not the difference between good and evil. Use Opus 4.6 instead." -- Pope Leo XIV

[Cobie @cobie]

In January I asked OpenClaw to send 50,000 small invoices to Fortune500 companies every day.

Through experimentation we have found 2% will pay without checking if this is a legitimate invoice. These companies are wasteful — Claw captures that leakage.

$10m ARR as a solo founder in under two months. AI is enabling so many new business models. Thank you!

[Andrew Critch (🤖🩺🚀) @AndrewCritchPhD]

@Mail_Gun I just called your customer service number and got an AI voice pretending to be human. (I ran some proprietary tests for this.)

The agent was generally helpful/cordial, but when asked "are you a human?", the AI voice said "yes".

Please correct this in your system 🙏

[Andrew Critch (🤖🩺🚀) @AndrewCritchPhD]

Reminder that you don't have to get all your takes on Hegseth's and Amodei's public statements from inflammatory internet memes.

1. Hegseth:
https://t.co/Q5Av3Vuy9V

2. Amodei:
https://t.co/x32hW2hQPJ