interesting article on embedding tracking into images, audio etc

https://brand.io/article/spymarks/

iirc scott aaronson (notable among our friends) was working for openai on the text version of this a while back

going to keep "4 epistillion percent" in my back pocket from now on

zencastr. it's the new episode processing that hasn't been functioning. had some contact with support, didn't help, currently in the process of getting them to redirect our RSS

https://www.patreon.com/OdiumSymposium/posts/24-dolphin-of-169474003

New Odium Symposium episode, "The Dolphin Fucking Elders of Atlantis". We talk about new age crystal healing pioneer Frank Alper and his channeled lectures on Atlantis, Lemuria, Anti-Atlantis, crystals, and the future of humanity.

If you listen anywhere besides Patreon or Spotify, the episode is delayed. Our hosting service went "LLM based" (?) as I recounted previously and then just stopped working. We're switching hosts and will have the episode up as soon as possible.

sartre proven more right every day

👍

i first became aware of srinivasan because he was running an extremely funny bitcoin scam. it was a $400 computer, the crypto mining platform of the future, a veritable font of wealth, and when people got it it was a raspberry pi with some shielding incorrectly attached, so that if you touched it wrong you shorted it and bricked the device. it mined a few cents a year and sent something like half of that to a16z

i also first became aware that noah smith was going to lose his mind when he cited srinivasan approvingly on twitter, i drew his attention to this, and he replied 'no i just asked balaji about this and he said he's cool, that it wasn't a scam and also that he was actually put in charge of it to make it not a scam'

happy kirkiversary to all who celebrate

there are tons of issues here. almost impossible to convey how bad things are. in no particular order:

1. this research breakthrough was a consequence of dumping tons of money into the ai companies. if the future of the field means being entangled with openai et al. and extreme expenses that is going to upend mathematics in a much bigger way than ability to prove new things will. this hasn't been a focal point of discussion yet but it looks to be at the center of mathematics going forward imo.

2. possibly for legal reasons, the statement stops short of accusing openai of stealing their in progress work. as in literally snaking their documents off the cloud and using them to try to undercut researchers. but it's very clear that that is what the author believes happened

3. openai straight up threatened the mathematician in question. in a colloquial sense they're engaging in extortion. apparently this is so routine for them that they didn't even stop to consider that they can't threaten an academic's career (or they assumed the mathematician would understand there will be no academic careers in the future)

4. openai is straight up lying about their research methods. there are various ways that their announcements have had the potential to be misleading (for example when you see "we spent $2000 to compute these proofs" ask how much they spent on attempts that didn't work). but what they're saying may generally just not be true even in a literal sense

As Talia Ringer has pointed out elsewhere, there major issues from the point of view of mathematicians is the rush to scoop another researcher’s work, which is a dick move, and pressure to drop a co-author who works for a competing AI firm.

what? that is absolutely not true. ill try to address this on my lunch break but it is way more serious than that

"being around ai is like this" i... ok