Mark Begemann markbe @markbe

Wild 24 hours for AI and lots of different proposals have been made. TLDR; the only *tangible* new fact is that OpenAI and Anthropic are going to have embedded 3rd party evaluators from unknown organizations with Dario floating METR as a possibility. Having 3rd party evaluators

471 849 6K 1.6M 3K

0 0 0 42 0

View Details

Mark Begemann markbe @markbe

a week ago

@myersfarm_ @elonmusk Apologies if you’ve addressed this previously (which is likely and I’ll look through your content.) Risk/reward of walking barefoot in pig manure? Don’t get me wrong, I thought nothing of stepping in cow poop yesterday… but I had shoes on. 😄

1 0 6 260 1

View Details

Mark Begemann markbe @markbe

a week ago

What is wrong with this argument? Reading comprehension being what it is today, please note that the argument is NOT that AI is safe. David Sacks has refined his position well.

David Sacks @DavidSacks

a week ago

Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead. You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier

3K 12K 72K 9.6M 15K

0 0 0 51 0

View Details

Mark Begemann markbe @markbe

a week ago

Fast and smooth down, push over the rollers for momentum maxxing, settle in just under pace and then slight negative splits in the middle, conserve energy during second set of rollers, pace to have nothing left at the top of the hill and almost die pushing over to full speed, let gravity and ego do the rest.

0 0 1 73 0

View Details

Mark Begemann markbe @markbe

2 weeks ago

@Brady_H I take 5-10g daily but after especially poor sleep, a 15g day seems to get me to a perceived normal-ish. Maybe just a good placebo but I’m rolling with it.

0 0 1 24 0

View Details

Mark Begemann markbe @markbe

2 weeks ago

Just tested all the major AI platforms free consumer versions with a simple task to find the best bacopa monnieri supplement. They all performed poorly and had to be led to the answer(s) while I endured their sycophancy. Weeeeak.

0 0 0 43 0

View Details

Mark Begemann markbe @markbe

2 weeks ago

@AutismCapital @jshchnz @OpenAI Most inspiring satire I’ve read in some time!

0 0 4 339 0

View Details

Mark Begemann markbe @markbe

3 weeks ago

Thomas Massie is naming names and I’m here for it.

Thomas Massie @RepThomasMassie

3 weeks ago

On the House floor tonight, I read the names of several Epstein coconspirators who should be investigated and prosecuted. I also began collecting 218 signatures for the discharge petition to force a vote on Epstein Files Transparency Act II.

5K 21K 93K 7.8M 5K

1 0 1 72 0

View Details

Mark Begemann markbe @markbe

3 weeks ago

@mikemo @thsottiaux How are you using the tokens, primarily?

0 0 0 25 0

View Details

Mark Begemann markbe @markbe

a month ago

Because it’s Terence Tao and lessons in math are applicable as patterns across many disciplines, obviously.

alphaXiv @askalphaxiv

a month ago

"Mathematics in the Age of AI" In an age of abundant AI-generated proofs, human understanding may become more valuable than proof generation itself. This new paper from Terence Tao argues that AI could make solving hard math problems much cheaper and faster, but at the same

40 213 957 53K 516

0 0 1 82 0

View Details

Mark Begemann markbe @markbe

a month ago

There is so much awesome stuff going on right now that my experience on X is to scroll my feed following builders until I get rate limited. Pretty stupid algo that can’t tell I’m real but then I don’t waste too much time here. Bye!