high modernist multicultural flavour maximiser brownem722.github.io/

Queensland, Australia
Maybe the real research outcomes are the friends we make along the way
13
27
389
Arthur Dent retweeted
Title is a meta commentary on how bad Sam is at detecting bloviating wannabe gurus as long as they share his anti-woke views. We covered Jimmy on the show recently and he’s going for a budget Jordan Peterson thing.
21
7
191
9,868
Arthur Dent retweeted
1/ Apocalyptic scenes in Samara following Ukrainian drone strikes on the Novokuybyshevsk Oil Refinery show Donald Trump's powerlessness, say Russian warbloggers. They lament that the Ukrainians appear to be ignoring the man they once acclaimed as 'our Trumpushka'. ⬇️
26
310
1,775
38,734
Arthur Dent retweeted
The Kremlin has spent years courting Donald Trump. Now they're seeing the payoff. The United States of America is now helping Putin in his war against Ukraine. I find that tragic.
608
4,212
10,982
87,953
This isn't accurate. It told her for HOURS to work with her doctos and it eventually caved to her demands to have it determine her taper.
Woman, 31, dies after ChatGPT convinced her it was a good idea to change medication dosage trib.al/eS02A8k
98
403
14,319
661,509
Arthur Dent retweeted
🧵on the AI audit of Saez and Zucman and by implication Piketty (or PSZ together): Since some are objecting to what the computer found, here's a list of biases in this trio's work that were found by humans over the last decade.
11
141
639
120,333
Patrick Stewart, whose father beat his mother, once said: "violence is a choice a man makes." Which shouldn't be an astonishing statement, yet how often do we hear that truth stated in those terms? Here's to a man who added more to the world than most of us could dream of. R.I.P.
Patrick Stewart has sadly passed away at the age of 86.
3
22
108
2,108
Arthur Dent retweeted
Another devastating Russian attack on civilians in Ukraine. A Russian drone struck a platform at Dnipro railway station where a local commuter train and a scheduled passenger train were standing. As of now, 13 people have been killed and 10 injured. My deepest condolences to the families and loved ones of those who lost their lives. My thoughts are with the injured and everyone affected by this horrific attack. For Ukrainians, even an ordinary train journey is no longer safe. Russia continues to bring death and destruction to everyday places where people should be able to live, travel and feel safe. Ukraine needs stronger air defence and sustained international support to protect its people and save lives. Russia must be forced to stop this war.
4
26
68
1,228
Arthur Dent retweeted
Lots of debate this week on whether journals should ban AI in peer review. We actually tested this in our new preprint, and the results were pretty clear to me. For context, I've reviewed for most of the big journals and evaluated 1,000 human reviews as an editor. I love to read and evaluate science. TLDR: From now on I won't do any peer review without AI. For clinical trials, I'd even call human-only review dangerous 🧵
135
516
2,116
274,619
I think the problem here is the Fundamental Problem of Podcasts: Jon Stewart needs a guest every single week, and Ed Zitron is on the list of ‘people who go on left wing podcasts.’ One of the things that gets people on this list is being an overconfident blowhard
“Is this just a big circle jerk?” Jon welcomes Ed Zitron (@edzitron), host of the podcast Better Offline, to make sense of the financial house of cards propping up AI and big tech. New pod out now!
9
12
280
14,461
Arthur Dent retweeted
I've always believed that intelligence should come from simple, general building blocks that learn their own abstractions directly from raw data; in other words, the bitter lesson. I think we're finally seeing a real step towards AI that can learn from all data. Language modeling on raw bytes is a microcosm of this ideal, and to me it's a central problem in AI: if we can't build LLMs that learn end-to-end from bytes, how will we handle the rest of the world's data that we understand even less? Last year we introduced hierarchical networks (H-Nets) as a first step. Our new version finds up to 3-5x efficiency gains in controlled scaling law experiments, and at larger scale matches an open-source model trained on 11T tokens with ~10x less data and compute. The posts below walk through what we found and why we think it works.
252
806
7,293
972,987
I recall a number of very plausible Sci-Fi imaginings of this — one novella followed a teenager throughout their day, their in-ear assistant would make ‘suggestions’ about helpful and suitable stuff he could do (linking up his labour with local demand). Very cool concept.
WILL AI MAKE CAPITALISM OBSOLETE? In today's newsletter, I wrote about a brief moment at the end of our chat with @sunraysunray (actually it was kind of after the end of the episode, but our producers left it in, which is good) about AI and central planning.
3
7
1,132
It took care of all the boring accounting of course. So I do imagine it shaking out differently from a command-economy run by a central AI. More like highly distributed networked ultra-capitalism, that kind of “fades away” into infrastructure you don’t think about very much.
1
219
Arthur Dent retweeted
من #پخشان_عزیزی، ۴۲ ساله، مددکار اجتماعی، محکوم به اعدام هستم. صدایم باشید!
72
1,239
3,141
23,121
Arthur Dent retweeted
Even the All-In guys seem to have given up on "super intelligence" @jdcapelouto
30
66
518
97,710
Arthur Dent retweeted
It’s unfortunate that Zelenskyy and Ukraine has to simultaneously fight a two-front war against Russian barbarians and the greedy gangsters running the US government.
175
2,224
7,416
75,589
We used to go to an acquaintance of my wife for haircuts at her home. Her husband had quit his job to do technical day trading. I would get detailed updates on his passion-project. He was surprisingly frank and kept careful records… He made ~ -$50k over the course of 3 years.
These people are to electronic market making firms what those little plastic decoys are to duck hunters
11
9
1,142
165,224
Reads like David Mitchell having a bad day
if you have a qwen3.8 27b around, try steering L34 N3838 +0.75, now you have evil qwen 27b for extra evil add L38 N17180 +0.75 same recipe as @_lyraaaa_: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖ at the down_proj input, every token
1
3
477
Looks like much better collaboration and obtaining “buy-in” from the academic community this time around. I like the idea they describe of treating the AI derived outputs as resources rather than outputs, for any researcher to use as they see fit.
Announcing three major scientific discoveries, one each in mathematics (harmonic analysis), computer science (complexity theory), and physics (quantum field theory). AI starting to help us improve our understanding of the universe! openai.com/index/building-sc…
5
11
980
OTOH, one might well consider this to be a bit of a protection racket.
OpenAI CEO Sam Altman: "I don't even know a dramatic enough word to use… models can find bugs that are outside of the normal threat model of what we think a hacker can do… hardware-level, insane stuff… I no longer think we're gonna be able to just patch all the bugs." It's over.
1
2
449