might be doing things // 2x exit // @PalantirTech // Alum @UMich

NYC
Joined August 2019
Wow. Reading some of @asmartbear 's (sold his co for millions, now CTO of a 9-figure biz) old blog posts and it's crazy how much of his wisdom is timeless. One of my favorite lessons: how quickly poor pricing can kill your startup:
2
10
did they default allow this 🤔
Branching of convos or "testing" an answer in LLM chats would be so useful. So so many times where I want to explore what would the LLM say if I prompted it one thing, but not lose my latest answer
I need someone who isn’t an angel investor to explain what the hell this means
BREAKING: We just raised $100M at @eightsleep to build the AI that finally fixes sleep. For decades, sleep has been passive. You lie down, close your eyes, and hope for the best. We’re changing that. Over 1B hours of biometric sleep data → distilled into fully autonomous AI that knows your body better than you do. We will do this in two ways: Step 1: Your personal Sleep Agent, running infinite nightly scenarios before they happen to deliver the most restorative sleep your body is capable of. Step 2: Thousands of Longevity Twins simulating potential health outcomes based on your data to help you live a longer and healthier life. Eight Sleep is becoming the operating system for sleep and longevity. Human life will never be the same again. Watch my full interview with @andrewrsorkin on @SquawkCNBC here
Zingers like this are stupid and weak. Obviously there will always be outliers, but saying it doesn’t matter is disingenuous as fuck The average person from Harvard will for fucking sure be more successful than the average state schooler
Replying to @felpix_
I got rejected from every school except Berkeley. I was also rejected from YC. It simply doesn’t matter. Just make things and build your own reputation.
1
my hot take is that in a lot of cases AI makes programmers lazier (and equally productive) but not faster
so all that money pouring into claude code is just a high end placebo? how is the collective this miscalibrated? this might be one of the most cognitively dissonant things i’ve consumed in a while.
The grok who visited my profile is fake af btw
For generating content this is amazing but when you're trying to really extract insights from these things it annoys the fuck out of me. Is there a paper out there that shows different prompts actually activate different knowledge circuits or is it all just verbal frosting?
is this a quirk of models that over-rely on web search? am i wrong?
Trying to prompt better to get advice from an LLM feels like a scam. It's operating on the same knowledge and feeding you different responses based on what you ask. Eventually you realize it's you talking to yourself, getting drunk off of self-confirmation bias.
you know what would be sick to build habits - get a group of 100 ppl - each person pays $20/m - at EOM raffle off money to 10 ppl - BUT you're only eligible if you post evidence of you doing __ habit every day for the month Keep some % of pool if you want to make it a biz
the scam veo3 ads are already out on youtube and are creepy as fuck
at least claude apologizes for being an idiot, o3 just gaslights you like it was right all along. fix @sama
o3 hallucinations make me want to punch a wall
1
1
“Urgent but not important tends to win over Important but not urgent” another banger by @asmartbear
1
1
o4-mini doesn't say shit but is levels ahead of sonnet and gemini pro
If you have experience coding, you can do this in <15 hours. Including timezone math.
if you think you can build calendly inhouse with 15 hours of prompting and vibe coding you have no idea how software works
1
Ronit retweeted
THE MOTORCADE IS HEADING TO THE PLAYOFFS 🏁 The Pistons will make their first playoff appearance since 2019‼️
It is so hard to get GPT-4.0 to generate me images in the correct dimension. It just seems to follow my instructions like 20% of the time. Super annoying.
Six diets were evaluated for their effect on speed of aging, plant-based and health-conscious diets slow aging the most, while fast food accelerates it. Diet defines your speed of aging, even in our early 20s. Genetics also play a big role in modulating diet’s effect on biological aging. Published today, a report shared the results of nutritional tracking of 413 twins (n=826, aged 21-25, 58.2% females) from the famous FinnTwin study. Based on detailed self-reporting a data-driven statistical model identified 6 nutritional patterns among participants (instead of theoretically assuming dietary patterns, a data-driven approach was used to cluster participants into patterns). Dietary patterns identified: 1. High fast-food, low fruits and vegetables 2. Plant based 3. Health conscious 4. Western with infrequent fish 5. Western with fish 6. Balanced average of all The effects of each the dietary patterns were compared population wise, and then within-paid analysis for 363 twin pairs (206 dizygotic, and 157 monozygotic) compared pairwise for the effect of their dietary choices and patterns on their biological age (GrimAge clock) and speed of aging (PACE clock) Twins are a precious clinical data mine, since they make perfect genetic controls on the case of monozygotic (real) twins, dizygotic (fraternal) twins share only 50% of their genomes (similar to normal siblings) but still offer the perfect control for all early environmental exposure in the womb and normally through childhood and adolescence. Comparing the effects in monozygotic vs dizygotic twins is a great tool to determine the genetic contribution to the phenomenon being studied. Three statistical models were used to evaluate the data, with increasing stringency of eliminating confounding variables: Model 1: adjusted for family relatedness, total nonalcoholic energy intake, and sex Model 2: Model 1 + smoking status and alcohol intake Model 3: Model 2 + BMI and the Baecke sport index Findings: i. The plant based diet was associated with highest reductions in the speed of aging, both on an entire population basis and in the pairwise comparisons of mono- and di-zygotic twins. ii. All dietary patterns also reduced the speed of aging compared to the high fast food low fruits and vegetables diet. iii. Plant-based diet association with a reduced PACE score held across all 3 models, with gradual decreases in PACE reduction 0.134, 0.109,and 0.08 for models 1,2, and respectively. This pattern held across most diets, with few exceptions. iv. In pair-wise comparisons between twins, plant-based and health-conscious dietary patterns were associated with the highest speed of aging reductions in both monozygotic and dizygotic twins, across all three confounder adjustment models. All other patterns reduced the PACE score in twin comparisons as well, similar to the whole-cohort analysis. v. Plant based and health-conscious patterns associated with slightly larger speed of aging (PACE score) reductions in monozygotic compared to dizygotic twins, unlike all other patterns, whose effects dwindled in monozygotic twins. This could indicate that unlike other diets plant based and health-conscious dietary patterns can reduce the speed of aging in a more genetically-independent manner than other diets. (caveat: the study authors warned from over interpreting the results with the twin-pairs analysis due to the small number of twins with distinct dietary patterns, the plant-based group had particularly a smaller number, which might have affected the results) The effect of the diets on biological age (GrimAge clock) was also shown in the study, general patterns were comparable, one difference was that the health-conscious pattern exceeded the plant-based in reducing biological age. We focused on the speed of aging in our analysis here due to two reasons i) the younger age of the participants, ii) the lack of a follow-up in the study.