Piplup
@spamxpip
Followers
15
Following
22
Media
13
Statuses
170
moderately hot takes on ai safety
Joined December 2023
what the hell is this thread 😭😭😭😭
You might remember Project Vend: an experiment where we (and our partners at @andonlabs) had Claude run a shop in our San Francisco office. After a rough start, the business is doing better. Mostly.
0
0
0
omg I found it at last
0
0
1
this is probably the funniest thread I've ever read lmao
@futureiscome @OpenAI Who the fuck wants companionship from an LLm?
0
0
1
is this LLM wrapper slop from zoom
Zoom achieved a new state-of-the-art (SOTA) result on Humanity’s Last Exam (HLE): 48.1% — outperforming other AI models with a 2.3% jump over the previous SOTA. ✨ HLE is one of the most rigorous tests in AI, built to measure real expert-level knowledge and deep reasoning across
0
0
0
me when I am an invigilator for a course that I cheated on
Proctoring an exam you cheated on has been the ultimate grad school experience
0
0
0
funnily enough all the research combining multi agents for a particular engineering task is probably the worst thing ever I need to write a position blog or something about this cause too many researchers are doing this slop and it's actually fucking this field up
0
0
1
I saw a company at NeurIPS this year selling wrong code for free to frontier labs r u guys ok that sort of data is worth a lot more than correct code
0
0
1
"you're in a ml engineer interview at blah blah inc" mannnn sybau and give the tldr in the first tweet if u want better engagement
0
0
1
idk man I went to NeurIPS for the research and instead all the research was slop but I did get really great connects tho
0
0
2
no clue why this happened I just received death threats and was gonna increase scores lol jk
ICLR prematurely terminated the reviewer’s edit score and comments. I have to ask: is this fair to those who haven’t finished their rebuttal yet?
0
0
1
PSA: If anyone is glaring at you at #NeurIPS2025, assume that they crawled the Openreview leak and recognize you as their Reviewer 2. Promptly report them to ICLR.💡 #AIAphorisms
5
8
85
I can predict next week's headlines: reviewers at NeurIPS assaulted due to openreview leaks
1
0
1
normalize researchers including ideas which did not work in their papers
1
0
2
we need polymarket to bet how long it takes for @elder_plinius to pwn opus 4.5 now
1
0
5
this pisses me off can someone just be biased to me and give me priority
0
0
1