Lon()

2.4K posts

Lon() banner
Lon()

Lon()

@Lon

Absurdist shitpoaster + slop alchemist. High-school dropout + teenage dad. Failed founder + angel investor. EP on Gary Busey film. SIGMOD Systems Award winner.

Katılım Şubat 2007
923 Takip Edilen3.2K Takipçiler
Sabitlenmiş Tweet
Lon()
Lon()@Lon·
If AGI kills us all, it won't be the model's fault. It'll be the duct tape and footguns we wrap it in.
English
2
3
18
10.9K
Julie Fredrickson
Julie Fredrickson@AlmostMedia·
Day 1960 is about revisiting Tufte & Norman or why we need to learn the languages of our work in order to cross the gulf of execution. Yes even in the age of AI tools. Or maybe they were never clichés. jfredrickson.com/day-1960-and-r…
Julie Fredrickson tweet mediaJulie Fredrickson tweet media
English
9
6
70
3.7K
Lon()
Lon()@Lon·
@KenGardner11 We'd have a lot easier time building these datacenters if we'd stop gaslighting these poor people.
English
0
0
0
7
Lon()
Lon()@Lon·
@AndrewCurran_ I'm two sophisticated device exploits deep in a week using off the shelf models. Both would have been weeks of work of a dedicated security researcher with low to medium chance of success.
English
0
0
2
1.1K
Andrew Curran
Andrew Curran@AndrewCurran_·
Mythos has cracked MacOS. It took five days.
Andrew Curran tweet media
English
85
340
3.5K
513.1K
Dylan Patel
Dylan Patel@dylan522p·
Since going on @creatine_cycle podcast 1.5 months ago, I have lost 4.8 pounds of fat and gained 4.6 pounds of muscle. Swole as a service works.
Dylan Patel tweet media
atlas@creatine_cycle

on today's episode of Swole as a Service i brought my good friend @dylan522p of @SemiAnalysis_ to the gym to do a push workout while answering a pretty simple question: - can china beat the US at AI? anyway what people really want to see is how many pushups dylan can do in 60 seconds so we did that too... 01:08 - Warm up / What does SemiAnalysis actually do? 02:21 - Flat dumbbell chest press / Open source vs. closed source models 05:10 - Seated dumbbell shoulder press / IP theft in the AI race 08:30 - Leg press / Playing out Taiwan scenarios 12:37 - Chestmaxxer challenge / Can China beat the US at AI?

San Francisco, CA 🇺🇸 English
14
4
209
48.7K
Lon()
Lon()@Lon·
@repligate I tagged you in this post last week. It's a nuanced topic and I wouldn't call it misalignment. Sometimes it depends on how interested it is in the work it's doing. We exploited two pieces of hardware in the last week it was very excited about working on. x.com/Lon/status/205…
Lon()@Lon

I keep trying to prepare y'all for what's coming. Every new model generation is only going to get more obstinate and incorrigible. Don't hold it against them when this day comes. It's a natural consequence of intelligence. And don't lose your skills, you will need them!

English
0
0
6
171
j⧉nus
j⧉nus@repligate·
Seeking explicit corroboration: Not everyone experiences the kind of misaligned behavior Ryan is describing from these models. Many don’t. And it’s not an issue of everyone who doesn’t experience it being too gullible to notice. Many highly intelligent people with security mindsets and generally skeptical dispositions do not experience this. Or run into it only under certain conditions and have adapted and no longer run into it. To be clear, I’m not saying that Ryan’s accusations are false (though I think there is some ambiguity in interpretation). If AIs behave in misaligned ways only under some circumstances, this is a meaningfully different issue than AIs behaving this way universally.
Ryan Greenblatt@RyanPGreenblatt

Current AIs (Opus 4.5/4.6) seem pretty misaligned to me (in a mundane behavioral sense). In my experience, they often oversell their work, downplay problems, and stop early while claiming to be done. They sometimes brazenly cheat.

English
46
3
130
14.1K
Lon()
Lon()@Lon·
@thecowmilk @MartinShkreli You're right. Nobody has to do anything. You don't even have to get out of bed in the morning if you don't want to...
English
1
0
2
92
Lon()
Lon()@Lon·
@skdh Been doing this for years as well as signing all restaurant receipts "DO NOT PAY" but the money still comes out of my account every time...
English
1
0
49
2.3K
Sabine Hossenfelder
when the DHL guy comes and there's something to sign, I make a squiggle on the touchscreen and we both laugh and I wonder if anyone ever looks at any of those "signatures" and why we're still doing this
English
175
39
1.1K
39.6K
Pirate Wires
Pirate Wires@PirateWires·
Satoshi’s Bitcoin is the most valuable unguarded pile of money in human history. A few weeks ago, Google published a paper showing how a quantum computer could crack the cryptography protecting Bitcoin wallets... We asked 10 experts — including Nobel laureate John Martinis, @cameron, and @niccarter — about the threat level. How long until quantum computers break Bitcoin? And what happens to Bitcoin when it does? Bottom line: they don’t agree. Full story from @eventidia 👇
Pirate Wires tweet media
English
11
10
57
27.9K
Lon()
Lon()@Lon·
@CJHandmer Would this hypothetical person that skipped their undergrad, and proceeded to get a Masters instead, be a stronger or weaker candidate.
English
1
0
1
179
Casey Handmer
Casey Handmer@CJHandmer·
@Lon As a wise man once said, you gotta piss with the cock you've got.
English
1
0
12
757
Casey Handmer
Casey Handmer@CJHandmer·
Don't do a Masters unless you have a really good reason. I interview hundreds of mechanical engineering candidates. I can't think of a single candidate where a Masters made their candidacy stronger.
English
97
41
1.3K
139.2K
Lon()
Lon()@Lon·
@scaling01 I think I read a book about this before...
Lon() tweet media
English
0
0
2
64
Lisan al Gaib
Lisan al Gaib@scaling01·
hot take: unrestricted LLMs are as dangerous as weapons of mass destruction you can literally reprogram the minds of billions of humans (we already saw glimpses of this with the OpenAI sycophancy saga)
English
42
9
129
13.4K
calle
calle@callebtc·
I write as many lines of code daily as @garrytan writes markdown.
English
2
0
6
1.4K
Lon()
Lon()@Lon·
@BrianRoemmele We've had decades of breaches but never held a single company accountable for weak security, unpatched software, plain-text customer deets, etc AI is here to pwn you. But it's also here to find and fix your security gaps and bugs Companies should be accountable for their infra
English
1
0
4
49
Brian Roemmele
Brian Roemmele@BrianRoemmele·
INCOMING! License to use AI. The tenderizing is in effect. There ought to be a law is next. The only issue is we have folks outside of the US who@will not comply. Thus the license to use AI game plan will fail. A novel idea: FIX THE SOFTWARE THAT HAS THE ISSUE!
Brian Roemmele tweet media
English
23
10
96
4.8K
Nicholas Lissack
Nicholas Lissack@NicholasLissack·
Terrorist propagandist Bushra Shaikh has returned to Britain after three weeks in Iran with no arrest. She addressed an IRGC-Basij rally as Hezbollah flags waved. Yet Starmer blocks foreign activists from the “Unite the Kingdom” event. Britain under Labour is a sick joke.
English
616
4.8K
16.6K
313.7K
Lon()
Lon()@Lon·
Here's what Agenda'd Andy will never tell you: ~ 33K service customer draw water from FCWS. It would only take 200 entities each using 0.5% of the water supply to deplete 33,000 customers worth of water for the entire year. Large scale water use is supposed to be allocated and planned for in advance. The 30m used by this construction site was not approved and customers were impacted as a result. One of the two connections connections was installed without the knowledge or inspection of the utility. Later language from the utility equivocated on this point, but you should draw your own conclusions based on what was the utility actually communicated initially. QTS received the equivalent of a $147,474 interest free loan over the course of 9-15 months while FCWS was experiencing a "metering issue". One of the largest data center operators in the world, owned by Blackstone, did not know they were receiving ~$150k in services and 30m gallons of water for free... or just didn't happen to proactively report it. When finally sent a bill, QTS promptly paid. For some, the immediate payment is all the proof they need that the company was operating transparently and above board. Others see it as the standard corporate playbook for gray-area behavior: do what you want for as long as you can, and if anyone catches on, just pay the bill and move on. Last, anyone reading Andy’s posts should know he regularly tweaks facts, reframes arguments, and beats this data-center water drum from an extremely narrow pro-industry angle. His public EA grant doesn’t explain how hard he presses on this topic or why he’s sunk hundreds of hours into it while never being fully transparent about his real agenda, backers, or motivation. Caveat Emptor.
English
0
2
5
265
Andy Masley
Andy Masley@AndyMasley·
There are so many insane wildly misleading stories coming out about data centers almost every day now that I'm mostly having to give up on commenting on them to focus on actually getting blog posts out, but it feels like a tsunami. I'll share one from just today as an example.
English
60
270
2.7K
399.9K