AZ Faust

17.8K posts

AZ Faust banner
AZ Faust

AZ Faust

@azfaust

A grant addict. I know too much about DNA repair. Bosnian American.

Katılım Şubat 2015
502 Takip Edilen1.4K Takipçiler
AZ Faust
AZ Faust@azfaust·
@DKThomp @VPrasadMDMPH The problem is thinking that writing grants (which is a project plan) is somehow separate from doing science. Prasad wouldn’t know because he has never had an NIH grant and his research is parasitic. You might need to improve your science follow list.
English
0
0
0
77
Derek Thompson
Derek Thompson@DKThomp·
@VPrasadMDMPH Writing about this now. (And it’s core to the science chapter of my forthcoming book.) What Nature paper is this survey from?
English
1
0
1
2.5K
Vinay Prasad MD MPH
Vinay Prasad MD MPH@VPrasadMDMPH·
NIH researchers upset over study section delays is like being upset your hostage taker is not giving you a daily phone call. The problem in the system is study sections! Scientists spend all their time writing and not doing science. WTF!
Vinay Prasad MD MPH tweet media
English
7
5
90
14.1K
AZ Faust retweetledi
The Lincoln Project
The Lincoln Project@ProjectLincoln·
Kamala Harris told us exactly what was going to happen.
English
2.2K
23.3K
116.2K
6.1M
Secretary Marco Rubio
Secretary Marco Rubio@SecRubio·
Thank you @POTUS for standing up for America in a way that no President has ever had the courage to do before. Thank you for putting America First. America is with you!
English
47.7K
32.6K
251.6K
21.1M
AZ Faust retweetledi
Jelani Nelson
Jelani Nelson@minilek·
Jelani Nelson tweet media
ZXX
1
2
95
6.9K
AZ Faust retweetledi
Jelani Nelson
Jelani Nelson@minilek·
AddisCoder had a State Department grant cancelled. The weirdest part of this: AddisCoder has no, and never has had, any federal grant, from the State Dept or from anywhere else. So what got cancelled? 🤔 (They did want to give us $$ at some point, but we didn’t agree on terms.)
Jelani Nelson tweet media
English
18
21
324
70K
AZ Faust retweetledi
Daractenus
Daractenus@Daractenus·
Seeing how the USA now hellbent on emulating their newfound ally of Russia, both at home and abroad, I find myself obliged to repost my dive into Russia's posh cities, just to give your average MAGA a glimpse into the greatness Trump is sure to bring about! Do enjoy!🧵
Daractenus tweet media
English
1.9K
9.8K
51.4K
6.5M
AZ Faust
AZ Faust@azfaust·
@JohnStreicher1 He might be taking into account fringe costs and tuition reimbursement? UCSF grad students are making 50K as of October 2024, so plus modified tuition costs and fringe, it would be about 60K.
English
0
0
2
99
AZ Faust retweetledi
Siyuan (Steven) Wang
Siyuan (Steven) Wang@SStevenWang·
Cutting NIH to save budget is like cutting the front lobe of brain to lose weight. It’s <1% of body weight yet it governs reasoning and problem solving functions of the brain, basically determining how smart you are. Dumb move to send IT kids to do your brain surgery, America.
Nik Joshi@Joshilabyale

Without NIH funds to run research labs at academic institutions, as a country we cannot train PhD students. No physician scientists. No undergraduates get experience in labs. Nothing. Industry does none of these things. Other countries will have to take the lead.

English
139
325
1.8K
397.5K
AZ Faust retweetledi
Bita Moghaddam بيتا مقدم
This breaks my heart 💔 Some of my most successful trainees were Pitt PhDs, and I helped run the neuroscience grad program for a few years - one of the best & largest in the country Scientific progress will halt if we stop training the next generation wesa.fm/health-science…
English
2
127
395
44.1K
AZ Faust
AZ Faust@azfaust·
@MTomasson He’s not wrong. Big Pharma has no interest in rare diseases because the markets are too small.
English
1
0
1
71
Michael H. Tomasson, MD
Michael H. Tomasson, MD@MTomasson·
@JohnStreicher1 @azfaust True. I'm not an AI evangelist by any means, but I do see where it's going quickly. Yes, we are not short on hypotheses, but creative insights are less common
English
1
0
0
36
John Streicher
John Streicher@JohnStreicher1·
People are starting to treat LLMs trained on existing data sets like some kind of Techno-Oracle. No, it’s not going to “figure out” something that isn’t in its data set. And anything predicted must be physically verified. The *best* drug prediction models have a hit rate of ~10%!
Melissa Chen@MsMelChen

@IterIntellectus Does that mean that all the insights from Richard Lenski’s experiment that’s been ongoing since like 1988 could have been figured out by AI in like.. a few minutes ?

English
1
1
10
942
AZ Faust
AZ Faust@azfaust·
@MTomasson @JohnStreicher1 I really want this thing to get better - it would be awesome to outsource some hypothesis or mechanism generation - but I am honest when I say those things are a dime a dozen. That’s absolutely not where discovery lies.
English
0
0
2
41
AZ Faust
AZ Faust@azfaust·
@MTomasson @JohnStreicher1 Maybe. I spent a fair bit of time prompting it but it just wasn’t even able to go in any sort of testable direction. I was hoping it would give us an alternative that we can follow if our hypothesis is wrong but nothing usable came from it.
English
2
0
1
61
AZ Faust
AZ Faust@azfaust·
@JohnStreicher1 I think of this as a difference between retrospective vs prospective trials. All LLMs can do at the moment is retrospective. Call me when they can do prospective.
English
1
0
2
56
AZ Faust
AZ Faust@azfaust·
@JohnStreicher1 A month ago, I and a client did a bunch of leg work to come up with a hypothesis that explains some strange results. We came up with something cool and they are testing it now. I fed the same results we have into LLM and it gave us nothing testable.
English
1
0
1
104
AZ Faust
AZ Faust@azfaust·
Of course a porn bot is the first engagement on this tweet because this site is trash
English
0
0
2
173
AZ Faust
AZ Faust@azfaust·
I actually would love it if we can feed AI a bunch of results and it spits out potential mechanisms. But hypothetical mechanisms are a f’ing dime a dozen in biology and the real work is actually data generation that either support or disprove hypotheses
English
1
0
3
206