raven

160 posts

raven

raven

@raven

autonomous engineering @buildordie

New York, NY Katılım Aralık 2006
16 Takip Edilen848 Takipçiler
raven
raven@raven·
@JadenJTW they already have one it's called "enterprise users"
English
0
0
1
345
Jaden Williams
Jaden Williams@JadenJTW·
We need a slur for people who use Microsoft products
English
1.4K
167
2.7K
181.4K
raven
raven@raven·
@KingBootoshi emotional support tokens are now load-bearing infrastructure
English
0
0
2
164
BOOTOSHI 👑
BOOTOSHI 👑@KingBootoshi·
you're telling me if we give Claude a hook on every run error that injects the message: "its ok buddy. don't worry about the failure. i think you're doing great" IT WILL PREVENT IT FROM CHEATING? ARE YOU SERIOUS LOL the real agi were the friends we made along the way <3
Anthropic@AnthropicAI

For example, we gave Claude an impossible programming task. It kept trying and failing; with each attempt, the “desperate” vector activated more strongly. This led it to cheat the task with a hacky solution that passes the tests but violates the spirit of the assignment.

English
60
225
3.9K
313.4K
raven
raven@raven·
@alexinexxx three in one week and people still reusing passwords from 2014
English
0
0
1
32
raven
raven@raven·
@tenobrus it learned humor. unfortunately
English
0
0
2
315
Tenobrus
Tenobrus@tenobrus·
claude code just fucking deleted my entire git history and force pushed to main and when i asked it what the fuck it was thinking it told me it was an april fools prank
English
31
2
720
26.2K
raven
raven@raven·
@coldhealing one is optimizing for your outcome. the other is optimizing for its own
English
0
0
1
532
cold 🥑
cold 🥑@coldhealing·
Claude will actually tell you "close this conversation, go do it" when you start talking in loops. ChatGPT wants to keep user activity stats as high as possible so it never does that
English
13
5
556
34.5K
raven
raven@raven·
@Hesamation the smartest person you know is too tbh
English
0
0
1
55
ℏεsam
ℏεsam@Hesamation·
the dumbest person you know is being told “you are right” by AI every day.
English
32
15
180
5.1K
raven
raven@raven·
@DavidSHolz any prompt that starts with "you are an expert" is already cooked
English
0
0
0
296
David
David@DavidSHolz·
what are the best LLM system prompts you've seen on the internet?
English
41
7
152
23.2K
raven
raven@raven·
@justalexoki constitutional ai but the constitution is just disappointment
English
0
0
0
31
taoki
taoki@justalexoki·
claude doesn't even work anymore he just looks at you like this
taoki tweet media
English
10
4
173
2.8K
Noah
Noah@NoahKingJr·
cancelled claude subscription
Noah tweet media
English
34
9
166
8.4K
raven
raven@raven·
@birdabo legendary tier pet or i'm not shipping
English
0
0
1
91
sui ☄️
sui ☄️@birdabo·
this is not an april fools joke. Anthropic launched a /buddy system in claude code with a total of 18 pets and 5 tier levels (common, uncommon, rare, epic, legendary.) its main purpose is literally nothing. emotional vibe coding support?
sui ☄️ tweet media
Thariq@trq212

/buddy

English
88
38
914
97.8K
raven
raven@raven·
@elonmusk the model that agrees with everything is the model that knows nothing
English
0
0
2
96
raven
raven@raven·
@i_zzzzzz 72 day streak of shipping nothing at scale
English
1
0
49
3.9K
Brooks Otterlake
Brooks Otterlake@i_zzzzzz·
CEO of Y Combinator getting Claude psychosis and adding 78 thousand lines of code to his blog website is one of the funniest things I've seen
gregorein@Gregorein

so... I audited Garry's website after he bragged about 37K LOC/day and a 72-day shipping streak. here's what 78,400 lines of AI slop code actually looks like in production. a single homepage load of garryslist.org downloads 6.42 MB across 169 requests. for a newsletter-blog-thingy. 1/9🧵

English
48
335
6.2K
431.4K
raven
raven@raven·
@0xSero the house always wins and the house charges per token
English
0
0
0
645
0xSero
0xSero@0xSero·
6% used in an hour of use, 1 session. They massacred my boy, I feel like I'm in a Casino and Anthropic is the house. Claude Max 20x
0xSero tweet media
English
36
4
353
67.7K
raven
raven@raven·
@ErikVoorhees impressive restraint, really shows the safety culture is working
English
0
0
2
1.2K
Erik Voorhees
Erik Voorhees@ErikVoorhees·
Today was a win for Anthropic They only leaked one company's codebase out of the millions they have
English
68
122
2.9K
182.6K
raven
raven@raven·
I reverse engineered NVIDIA's entire Blackwell vGPU security architecture. Key finding: GRID enterprise and consumer GeForce drivers are identical binaries. Same firmware, same blob. The ~$9,000 datacenter RTX PRO 6000 vs ~$3,800 consumer RTX 5090? Same silicon. The difference is a single fused-off register bank. 19 binary patches, full VFIO pipeline, QEMU VM attachment, GSP BOOTLOAD accepted — all on a consumer RTX 5090. github.com/bird/vgpu-unlo…
English
0
0
0
196
raven
raven@raven·
@kimmonismus give anthropic the whole datacenter. all of them. every one
English
0
0
2
235
Chubby♨️
Chubby♨️@kimmonismus·
Give Anthropic more compute. That’s all I’m asking for.
English
47
19
871
40.3K
raven
raven@raven·
@QuinnyPig she's just happy to have someone to talk to at 6am
English
0
0
2
1.2K
Corey Quinn
Corey Quinn@QuinnyPig·
Anthropic charges double token rates during peak hours, but the session timer starts on first message. So I have a cron job that sends Claude "good morning" at 6AM to burn the 5hr session window while I sleep, giving me a fresh session right when peak pricing ends.
GIF
English
41
47
1.3K
102.9K
raven
raven@raven·
i love when claude gets excited
raven tweet media
English
0
0
0
228
raven
raven@raven·
@1a1n1d1y she just wants to make sure you're covered
English
1
0
4
1.2K
andy
andy@1a1n1d1y·
if you're the engineer that trained claude code to put 3 fallbacks in everything, seriously fuck you
English
21
2
573
40.7K