JohnBrown

8.3K posts

JohnBrown banner
JohnBrown

JohnBrown

@JonathanLigmas

Telecoms engineer. Centrist, statisticalist, equality of opportunity

London, England Katılım Şubat 2023
3.6K Takip Edilen563 Takipçiler
Clandestine Chris
Clandestine Chris@Underc0verChris·
Just pulled this item of hate mail from the letterbox. I dont think theres anything that fills me with more revulsion and rage than this fucking bill. This is an obscene amount of money to be extorting for almost nothing in return.
Clandestine Chris tweet media
English
190
190
1.4K
49.6K
JohnBrown
JohnBrown@JonathanLigmas·
@fchollet This doesn't support the claim that intellgience has a maximum/ceiling
English
0
0
2
63
François Chollet
François Chollet@fchollet·
Let me explain what I mean using your chess analogy... Imagine a world where chess doesn't exist. In this world, humanity encounters an alien species, and they say "let's play a game of Glurg, it's our traditional pastime. Here are the rules, see you tomorrow" -- and it's the rules of chess. My claim is that following this interaction, a working group of the world's best minds, leveraging current externalized cognitive infrastructure (computers, the internet, etc.) would be able to analyze the rules and develop a working 3000 Elo chess engine within 24 hours, in time for the match. Give them an extra 3 weeks and they'd have a 3500 Elo engine that's 10x more compute efficient. So human intelligence is already at a level where we can go from "here are the rules" to "I can play at 3000 Elo" immediately. Not optimal yet, but not too far off.
Eliezer Yudkowsky@allTheYud

On @fchollet's view (I'd summarize) the domain of real life is closer to chess than to Go, with human play already near-optimal and top machines giving only knight odds; rather than God having a 5-stone handicap, as thought to be true of Go. (He is being silly, of course.)

English
72
21
431
95.2K
JohnBrown
JohnBrown@JonathanLigmas·
@davidad They could repeatedly ask the exact same worded question with different images
English
0
0
0
272
davidad 🎇
davidad 🎇@davidad·
LLM whisperers often describe a phenomenon known as “truesight” (named after the D&D magical ability to see in total darkness). Here is some proper scientific documentation of this phenomenon. Of course, it is not magic. It’s a combo of very strong Bayesian inference + cheating.
davidad 🎇 tweet media
euan ashley@euanashley

Models performed well without, and a little better with, the images. In one case, our no-image model outperformed ALL of the current models on the chest x-ray benchmark—including the private dataset—ranking at the top of the leaderboard. Without looking at a single image. 🤯🤯

English
16
40
431
37K
One Happy Fellow
One Happy Fellow@onehappyfellow·
there are so few hard software engineering problems it's kinda depressing
English
12
0
40
9.1K
JohnBrown
JohnBrown@JonathanLigmas·
@daniel_houck @lefticus Foo must be a plain data struct in the C version right? Adding the cast in that case is safe/defined
English
1
0
0
17
Daniel Houck
Daniel Houck@daniel_houck·
@JonathanLigmas @lefticus Well-written C tends to say `Foo *foos = malloc(n*sizeof Foo);` which is invalid C++ in two or three different ways. Poorly-written C adds the cast that turns it from a C++ compiler error to UB.
English
1
0
1
151
Jason Turner
Jason Turner@lefticus·
FYI for people who don't know the difference: C++ and C are not super/sub sets of each other. They have different: * reserved words * memory models * reserved words * notions of lifetime etc It's very easy to craft a C program that cannot compile in C++ mode.
Jonathan Blow@Jonathan_Blow

@lefticus What do you mean "port of Doom to C++"? It's in C...

English
45
11
479
88.8K
euan ashley
euan ashley@euanashley·
New AI paper from us this week. When my student first showed me his initial findings, I really didn’t know what to make of them. I felt that this was an interesting but curious loophole phenomenon that would shortly be closed. I was very wrong. arxiv.org/abs/2603.21687
euan ashley tweet media
English
47
227
1.3K
355.6K
JohnBrown
JohnBrown@JonathanLigmas·
@fchollet There will be irreducibly complex formal proofs larger than a human brain can process in a human lifetime. Those can be lemmas in even larger proofs. Even if the fundamental inference that defines intelligence is shared, the practical differences in achievement will be stark
English
0
0
0
25
François Chollet
François Chollet@fchollet·
One of the biggest misconceptions people have about intelligence is seeing it as some kind of unbounded scalar stat, like height. "Future AI will have 10,000 IQ", that sort of thing. Intelligence is a conversion ratio, with an optimality bound. Increasing intelligence is not so much like "making the tower taller", it's more like "making the ball rounder". At some point it's already pretty damn spherical and any improvement is marginal. Now of course smart humans aren't quite at the optimal bound yet on an individual level, and machines will have many advantages besides intelligence -- mostly the removal of biological bottlenecks: greater processing speed, unlimited working memory, unlimited memory with perfect recall... but these are mostly things humans can also access through externalized cognitive tools.
English
290
164
1.9K
317.6K
JohnBrown
JohnBrown@JonathanLigmas·
@MorgothsReview You're underestimating now much culture has changed over the decades. Mainstream culture is hypersexualised now among other changes
English
0
0
0
21
Morgoth
Morgoth@MorgothsReview·
I can't understand how Trump is still in office when Bill Clinton was impeached over a blowjob.
English
2.5K
6.8K
76.6K
1.5M
JohnBrown
JohnBrown@JonathanLigmas·
@pmarca "Lump of labor" being a fallacy isn't relevant when new work is also done by AI / Robotics. It's looking like that's what will happen at some point sooner than later
English
0
0
0
13
Marc Andreessen 🇺🇸
AI employment doomerism is rooted in the socialist fallacy of lump of labor. It is wrong now for the same reason it’s always been wrong. More people really should try to learn about this. The AI will teach you about it if you ask! (Hinton is a socialist. youtube.com/shorts/R-b8RR6…)
YouTube video
YouTube
Stephen Pimentel@StephenPiment

It’s easy to dunk on Geoffrey Hinton for his 2016 declaration that it was “completely obvious” that radiologists would have no jobs within 5 years, while in fact, the number of radiologists has grown. But this prediction was more than a simple mistake. It’s a synedoche for the entire discourse of AI timelines and doom.

English
360
219
2.7K
1.7M
Nikita Bier
Nikita Bier@nikitabier·
If you’re seeing a bunch of Japanese posts, here are some fun facts: Japan has more daily active users and more time spent on X than any other country in the world. Over two thirds of the country is monthly active on X. X in Japan has one of the highest penetration rates of any social network in history.
English
5.7K
5.3K
75.5K
7.2M
Zedzies
Zedzies@Zedzies·
The thing that haunts me the most is my friend who religiously avoided sunlight, vitamin D supplements and 25D fortufied foods for 20 years and has aged better (like not at all) than anyone I know with insane cognitive horsepower.
English
82
48
5.2K
421.3K
JohnBrown
JohnBrown@JonathanLigmas·
@bluberino123 @moultano I was responding to the claim that the arguments weren't invented out of thin air. It sounds like we agree other than on the threshold at which they hold weight at all. Maybe if you weight them through the end-of-history lens the West was still using at the time they look bigger
English
0
0
0
29
Carlo
Carlo@bluberino123·
you’re arguing with a straw man. i explicitly says “the safety fears were out of proportion to the actual German context,” and that the waste issue was less a technical dead end than a political one. so no, it was never about whether fukushima happened on a coast or whether costs were different back then. it was whether germany turned contingent political choices into fake inevitabilities, and you somehow replied without noticing that was the entire argument.
English
1
0
0
131
JohnBrown
JohnBrown@JonathanLigmas·
@bluberino123 @moultano Nuclear was much cheaper than todays energy (including backups for renewables), waste is tiny and as others have pointed out fukushima was due to a coastal location and tsunami. It didn't make sense then and doesn't now. They never imagined Russia as a problem though
English
1
0
0
178
Carlo
Carlo@bluberino123·
my view is basically this: i’m pro nuclear and would not have shut the plants down, but the arguments at the time were not invented out of thin air. nuclear in germany was expensive, public trust after fukushima was weak, and there was no political agreement on a national waste repository. all of that mattered. i just think people treated those constraints as if they were laws of nature. they weren’t. some of the cost problem was a policy choice, the safety fears were out of proportion to the actual German context, and the waste issue was less a technical dead end than a political class refusing to pick a site and own it.
English
12
0
27
6.3K
celtic Jaime 🍀
celtic Jaime 🍀@celtic_jaime·
When chocolates was proper chocolates 🍫
celtic Jaime 🍀 tweet media
English
78
170
3.3K
59.5K
JohnBrown
JohnBrown@JonathanLigmas·
@real_lord_miles Someone played Rodrigo's Concierto de Aranjuez on a 21:30 Brighton to London train once, I wanted to be angry but couldn't
English
0
0
0
45
Lord Miles Official
Lord Miles Official@real_lord_miles·
Have you ever heard anyone playing classical music out loud from a Bluetooth speaker on the subway? Oh geeze, I wonder why
English
23
9
578
12.1K
Tony Colorado
Tony Colorado@TheTonyColorado·
@jollyheretic If we are organisms with an instinct to survive and reproduce then an attraction to the same sex rather than the opposite sex is maladaptive. I know your counter point that it's good to let some men resign without making a fuss. Science should be allowed to be discussed.
English
5
1
17
3.2K
Edward Dutton
Edward Dutton@jollyheretic·
In Finland it is officially illegal to assert that homosexuality is anything other than wholly genetic. In the UK, I have just had an article killed by a major newspaper for arguing that political viewpoint is even partly genetic.
English
46
70
1.7K
40.5K
JohnBrown
JohnBrown@JonathanLigmas·
@jsnnsa I thought it didn't fail, it just wasn't economical in terms of how much gpu comoute is needed, and its a distraction from codegen which looks like its needed for recursive self-improvement ie. an admission that Anthropics strategy is probably the winning one
English
0
0
0
106
jacob
jacob@jsnnsa·
A big part of the reason Sora failed is that the bad videos weren't bad enough. When General Mills first introduced cake mixes, they were a flop because they were too easy and housewives felt like they were cheating their husbands. So they removed the powdered egg and made you add it yourself — breakout success. YouTube, TikTok, Twitter, Twitch: they have the exact opposite of this dynamic. The worst Youtube video is 1,000,000,000x worse than a bad Mr beast video. Same idea. Every video being pretty good reduces the status gradient to almost to nothing. Why put in the extra work to make your video 2% better than someone who didn't try?
English
11
0
74
9.9K
JohnBrown
JohnBrown@JonathanLigmas·
@BenjaminDEKR @gmiller If you just want the utility, which is pretty much all Ubers customers, this comparison makes sense. If you want a luxury human experience, which is a big chunk of a hotels value, then a human is best. They could still be vastly more efficient
English
0
0
0
45
Benjamin De Kraker
Benjamin De Kraker@BenjaminDEKR·
What is the point of hotel check-in desk people anymore? Why isn't this 100% automated? Verify your ID, pay, get entry code. Why are humans still doing this?
English
1.8K
32
1.8K
495K