Sabitlenmiş Tweet
Sam Finlayson
9.9K posts
Sam Finlayson
@IAmSamFin
MD, PhD. Pediatrics-Genetics @seattlechildren and Medical AI @EvidenceOpen Past: @harvardmed, @MIT_CSAIL, @Stanford
Seattle, WA Katılım Ekim 2014
680 Takip Edilen5.3K Takipçiler
Not typically a fan of rage bait, but swapping isomorphic and orthogonal is genius and the reply guys are running in like rats to cheese
anirudh@kamathematic
just landed in sf? “update your priors” = update your world view “orthogonal” = analogous “isomorphic” = in no way similar “overfit” = overdressed “steelman” = devil’s advocate
English
This is the first major project at OE that bridges my two great academic passions of rare disease and AI, and the goal is for it not to be the last. And if you are interested in being an export involved in the project (time compensated), feel free to reach out. 4/4
English
@RareDiseases works very hard to curate info for individuals impact by rare disease, but building and maintaining these is laborious. With this project, we hope to supercharge their efforts, and surface this info to clinicians who often get limited exposure to these conditions 3/4
English
Excited to share this collaboration between @EvidenceOpen and @RareDiseases that I've been helping in part to spearhead (1/4):
rarediseases.org/national-organ…
English
Medical interns when the bots are finally good at charting.
taoki@justalexoki
doctors are about to get automated and i think literally everybody is happy about that
English
@DanielleFong Shouldn’ve been Betty
en.wikipedia.org/wiki/Betty_Sha…
English
i mean, do we have to say it
Jason Crawford@jasoncrawford
Um… Claude Code just created a .claire directory? For its git worktrees. Who is Claire?
English
Sam Finlayson retweetledi
@DeanTTraining A situation has arisen in which intellectual integrity may demand a menswear-guy-like action on your part... x.com/SecKennedy/sta…
Secretary Kennedy@SecKennedy
Leg day in Austin.
English
Something like 95% of all gym goers have never even THOUGHT about taking a rep this hard one single time let alone regularly
This is why 95% of gym goers go years and sometimes even decades without making ANY progress whatsoever
Leg days TEACH people what true intensity is
Rxchxrdx@deadliftblue
On social media you will see alot of people who can’t wait for leg day either aren’t training hard or have never seen God during rep 9 & above. A lot of lifters aren’t fearful of leg day because they never take a set into real struggle. I love leg day but I am a masochist
English
Sam Finlayson retweetledi
Nikita is joking (I think) but a lot of medium trust systems that relied on there being just enough friction to discourage minor fraud are about to break at scale.
Nikita Bier@nikitabier
My agent looked up every Amazon product I've bought in the last 10 years, called each manufacturer, said it broke and demanded a replacement. I now have 6 TVs, 12 printers, 2 microwaves, and 800 tubes of tooth paste.
English
@giffmana But what if dumber conscripted WarClaude just leads to hallucinated terrorism accusations and “you’re absolutely right —I killed the wrong Joe Smith!”
English
Holy cow! I hope for the sake of my friends there, that it won't happen... but Anthro -of all the labs!- being the one forced to train a WarClaude would be the most ironic outcome possible!
Btw, completely unrelated, but i noticed that if you set the learning rate just 3x too low, the model looks perfectly reasonable but it's just dumber across the board, and i haven't found a way to tell by looking at the final model artifact.
GIF
Andrew Curran@AndrewCurran_
Update on the meeting; according to Axios Defense Secretary Pete Hegseth gave Dario Amodei until Friday night to give the military unfettered access to Claude or face the consequences, which may even include invoking the Defense Production Act to force the training of a WarClaude
English
@tallsnail @moniqueo59 UK guidelines on the topic much better communicated than US ones, IMO
lullabytrust.org.uk/baby-safety/sa…
English
@moniqueo59 Also the "abstince-only" guidance a lot of new parents receive actually can drive more unsafe sleeping decisions out of desperation
x.com/i/status/20256…
lia@tallsnail
@Ivana331Mendez Yes. I personally know multiple people who have said to me "well the doctor said never bring the baby into my bed so i've been sleeping with the baby on the couch" 😑 so much more dangerous
English
seeing dozens of parents in my reddit due date group talk about how wrecked they are because their newborn "won’t sleep in the bassinet". taking night shifts, setting 15-minute alarms to blast themselves awake constantly so they don't doze off holding the baby
it makes me really sad
just sleep with the baby. sleep with the baby!! do it safely but just sleep with him. it is your birthright and his, too
English
Sam Finlayson retweetledi
The reality is that even with very large sample sizes (women: 130K cases/159K controls, men: 65K cases/132K controls), the study didn't find a single significant genotype-by-sex interaction.
The larger number of genome-wide hits in women is simply the result of more power.
nature@Nature
A comprehensive genome analysis suggests that women have more genetic risk variants for major depressive disorder than do men, with stronger links to metabolic traits go.nature.com/47rZnXR
English
@JamesADiao @JAMAInternalMed Congrats, man. Strong work as is typical for those involved!
English
1/ We’ve heard so much from researchers and academics on the tradeoffs of using race in medicine. But what do patients want?
We conducted the first nationally representative survey to systematically study these preferences, now out in @JAMAInternalMed:
English
@alexrkonrad @penzer_j Alas, C(MO)'s answer was itself clearly written by ChatGPT.
English
If you were in the OpenAI slack channel or groupchat that workshopped this "more Texans use ChatGPT than Claude has users" line, please DM
English
@BrandonLuuMD Love this feedback. The majority of OE questions are management questions, so I’m glad that’s a strength in your experience. We’ve been actively working on some tweaks to the differential system so hopefully that gets better soon.
English
My current AI medicine stack:
Differential diagnosis and investigations:
-UpToDate AI
Treatment:
-UpToDate AI
-OpenEvidence
Rare and complex diseases:
-Grok DeepSearch
-Perplexity
-OpenEvidence
Brandon Luu, MD@BrandonLuuMD
UpToDate’s new AI is a genuine game changer. Every output is clearly linked to the underlying primary articles. It’s also remarkably strong at generating accurate differentials. Easily the most reliable clinical AI tool I’ve used so far.
English
Sam Finlayson retweetledi
@mackinprof @drjohnm exercise is good
pick your favorite types
do it enough so that you are winded or sore a few times a week
don't be a weekend warrior
if its your cup of tea, go beyond that
but save your time & money doing epi research on exercise
English