directionally correct always | unqualified to pontificate | king of the low roll | your favorite midwits favorite midwit | swva
San Francisco, CA
- Tweets 4,772
- Following 757
- Followers 1,733
- Likes 24,221
> woke up giga early so i could walk in to the DMV
> get there iβm third in line yippee
> immediate announcement βall systems are down, we have no timetable for when itβll be backβ
fuck my chungus life
2
15
759
random thoughts on human data:
human data is fundamentally bottlenecked by good SPLs
a good SPL has meticulously gone through taste training, developing an intuition for which tokens are useful and which are slop
there is no substitute for experience here, the skill only compounds with # of projects done
as a result SPLs do and should command insane pricing power, one good SPL can unlock a huge multiple of lab data spend (compared to salary/bonus)
its no surprise that data co's who have not invested time training/upleveling SPLs have stagnated while those that do are up infinite
synthetic data might take over against human data but this may actually cause a jevon's paradox + more demand for SPLs
still early imo for useful token surveyors
7
3
135
13,359
my new meta for the 80/20 on system design is i stick my good friends devin, inspect, claude, + codex into a single slack thread and tell them to engage in a spirited intellectual debate until they agree on what to do
ofc i have to interrupt to calibrate and refine final output
5
514
sometimes you see a tweet on the timeline and you immediately think "wow this guy really doesnt know"
many such stories amongst the schizo posters
7
351
every business in existence has the ability to be a data company (theoretically)
I asked people: "which data companies should an ambitious future founder join?"
13 votes: Mercor (@BrendanFoody @adarsh_exe @suryamidha)
11 votes: Fleet (@nicoup @andrewthezhou)
10 votes: Proximal (@MatternJustus @navidkpr @calvinchen)
9 votes: Surge (@echen)
8 votes: AfterQuery (@CarlosGeorgescu @spencermateega)
7 votes: Mechanize (@GuiveAssadi @tamaybes @EgeErdil2 @MatthewJBar)
6 votes: micro1 (@aliansarinik)
4 votes: Datacurve (@serenaa_ge @charleyslee)
4 votes: OpenAI (data team) (@BarrAlexandra @phoebethacker)
4 votes: Preference Model (@Ning_Catsnail)
3 votes: Bespoke Labs (@madiator @AlexGDimakis)
2 votes: Anthropic (data team)
2 votes: Datology (@arimorcos @hurrycane @leavittron)
2 votes: Long Lake (@alextaubman @rasmuswissmann @varunshenoy_ @tarasholovko)
2 votes: Sieve (@mokshith @thpicy)
People couldn't pick a company they work at or founded. Disclosure: I'm a small investor in AfterQuery and Fleet (in this post), and in Abundant, Applied Compute, Latch, Olam Labs, Trajectory and Vmax (elsewhere on the list). I didn't vote.
13
866
one day i hope to have even a fraction of the aura agamemnon had in the odyssey movie when the gates of troy opened
5
407
tigs is jestermaxxing ππ³ retweeted
i'm judging @hackmit this weekend say hello if you're around!
2
2
50
4,081
if ppl you meet at events are still asking what college you went to, you might have to go to higher signal events π«¨
i have no desire to know where anyone went to school other than to talk shit about college football
1
12
967
shoutouts to @jesslynteojy + @LiuLawrence45 for setting me up with a @WillowVoiceAI !!! ready to put the debate spreading skills to good use
hit them up if u want a personalized onboarding
2
1
11
560
thinking about the ai alignment problem, there was a comic i saw last yr called "the suit" by scott base
basically, a human astronaut has an intelligent survival space suit with the only goal of keeping the human alive on his journey home
it recycles water, helps him march
but also recycles skin and even body parts for nutrition for the human
by the end, the human has his eyes recycled and essentially is just a consciousness with no senses whatsoever
even the most innocuous of directives to an ai can be interpreted in a way that might end with an "i have no mouth and i must scream" situation
generally believe alignment + safety should be quite important + high pri in the labs
5
628
openai ipo postponed for safety reasons
if anthropic is a real one theyβd do the same
sf housing market prices intact for one more year?
8
515
red teaming my single homie's instinct agent to ask his crush on a date
Introducing our Trusted Person network powered by our new Instinct-to-Instinct communication protocol. Your Instinct can now talk to other Instincts to coordinate plans on your behalf.
Getting people together often involves a lot of back-and-forth: finding a time, working out the details, and following up when plans change. Instincts can now work with each other to help handle that coordination.
For example:
- Coordinating plans with your spouse
- Planning a weekend trip with friends, including dates and arrival times
- Scheduling time with colleagues and other collaborators
- Hosting events with 10s or 100s of people
- Organizing recurring plans like tennis, book clubs, or weekly dinners
Disclaimer: Your Instinct will only be able to communicate with those in your Trusted Person network. This is the group of people who are closest to you:
- Spouse
- Older children and parents
- Colleagues and trusted professional relationships
- Small local businesses
The Trusted Person network is currently available to our early access group and will be slowly rolled out across the user base over the next 24-48 hours.
1
10
3,183
neolabs?
astra just solved navier stokes and can do everything from object detection to spatial reasoning to whatever tf else u need it to do
oai + anthropic both claim to have models internally that make astra look like gpt5.4
meta just dropped their personal ai agent and are apparently cooking up fable5.1 level intelligence ezpz
xai has TOO MUCH compute and also all the coding data in the world
google is doing stuff too most likely (sadly flash3.8 overshadowed so far by everything else but its probably sick)
if anything, its clear that the bitter lesson is very real -- compute and data are basically all you need
and you're still investing in neolabs?
7
109
12,329