Pavle Jeremić

727 posts

Pavle Jeremić banner
Pavle Jeremić

Pavle Jeremić

@SynBioMars

Building the post scarcity future

San Francisco, CA Katılım Şubat 2014
541 Takip Edilen391 Takipçiler
Sabitlenmiş Tweet
Pavle Jeremić
Pavle Jeremić@SynBioMars·
Some learnings from the last 12 months. At Aether our mission has always been to build a future of abundance for the human race. We plan on doing this by using our AI to design totally new classes of proteins that can assemble complex products in modular factories (more on this another time).
English
1
3
18
2.3K
Pavle Jeremić retweetledi
Daniel
Daniel@growing_daniel·
Make San Francisco a shitty place to be a drug addict
English
57
56
1.9K
54.1K
Pavle Jeremić retweetledi
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
Star Wars depicts a future where cybersecurity just doesn't work. They have AGI but they keep it bottled up in droids; they don't network anything. As soon as R2-D2 gets access to an actual network he successfully hacks the Death Star.
English
153
289
9.8K
289.5K
Pavle Jeremić retweetledi
Crémieux
Crémieux@cremieuxrecueil·
I've said it once, I'll say it again and keep repeating myself: No expensive housing market builds lots of housing. That is not a coincidence: build homes and prices are controlled!
Crémieux tweet media
Owen Zidar@omzidar

Striking graph from @esoltas and Jon Gruber

English
35
115
1.5K
63.5K
Pavle Jeremić retweetledi
Gianmatteo Costanza
Gianmatteo Costanza@emissionite·
Whenever the Big One hits San Francisco and levels vast swaths of the city, we will finally confront an uncomfortable truth: we have outlawed our own existence. Cities built over generations cannot be rebuilt because our codes are anti-city, and most people don’t know it.
Sukrit Ganesh 🇺🇸 🥑 🚲🛩️@SukritGanesh

Lahaina burned down in 2023. As of 2026, few homes have been rebuilt. One big reason why? Lahaina's zoning is one of the most restrictive in the US. Only low-density SFHs allowed. Most can hardly afford to rebuild. There is demand for new apartments, but the law prohibits it.

English
37
116
2.1K
89.3K
Pavle Jeremić
Pavle Jeremić@SynBioMars·
This is why all these startups have generally failed, algorithmic approaches for over 10 years have been able to find targets, but the real value unlock only happens well into clinical trials when safety and efficacy are determined!
Crémieux@cremieuxrecueil

Yep! We have so many drug targets. Finding targets is not the issue. The issue is testing them and going through the rigamarole to get them produced and on the market. AI-in-medicine people are missing this and talking about an area that just isn't the bottleneck to new drugs.

English
0
1
2
346
Pavle Jeremić
Pavle Jeremić@SynBioMars·
@Patrick_Maksoud @p_maverick_b Yeah, generated internally is the only way, but I’ve at least not seen anyone outside of Aether generating the right type (and volume) of data to train these models
English
0
0
0
38
Patrick Boyle — e/🦀
Patrick Boyle — e/🦀@p_maverick_b·
I've heard biologists complain that AI isn't useful because new bio data doesn't make the models better... but what if AI is a mirror showing us that a lot of data isn't as insightful as we think it is? How do we identify experiments that are maximally informative to a model?
English
19
14
155
11.7K
Eryney
Eryney@eryney_ok·
I remember when I was in college there was an obsession with gold nanoparticles. Like every chemistry journal was flooded with methods for synthesizing or using them, you would’ve thought it was gonna be a Nobel prize worthy discovery. But then it just like, stopped. I’m not a chemist, so I have no real insight for why this was or what happened. To complete the meme: You don’t hear much about gold nanoparticles anymore. Lesson in there somewhere.
Eryney tweet mediaEryney tweet media
Aashay Sanghvi@aashaysanghvi_

You don't hear much about 'nanotechnology' anymore

English
17
6
107
27.1K
Pavle Jeremić retweetledi
Kane 謝凱堯
Kane 謝凱堯@kane·
@saikatc Saikat, are you aware than San Francisco county has a $15.9B budget for under 900k people? We could halve the budget and still be spending double what Denver county spends. There is no funding shortage, just a lot of corruption.
English
17
56
1.9K
60.4K
davinci
davinci@leothecurious·
@SynBioMars @tszzl oh didn't know that. interesting. so capturing the structure of a real protein under natural conditions remains an unsolved engineering problem?
English
1
0
2
73
davinci
davinci@leothecurious·
over the span of 60 years, humans experimentally identified ~200,000 protein structures, with a single structure typically requiring 1-7 years of sustained work by a research team. in roughly 3 years, alphafold predicted over 200,000,000 sturctures, often close to experimental accuracy. structures are now documented for nearly every protein known to man, and future predictions mainly revolve around discovering entirely novel sequences. this is by far one of the most impressive examples of large-scale, highly consequenital cognitive labor automation in human history.
davinci@leothecurious

@tszzl alphafold is up there

English
29
143
2.4K
161.4K
Pavle Jeremić
Pavle Jeremić@SynBioMars·
@leothecurious @tszzl Put another way, the data that’s used to label the protein sequences with structure is unrepresentative of the true protein structure, which is what could lead to understanding the physics behind it
English
1
0
6
104
Pavle Jeremić
Pavle Jeremić@SynBioMars·
The issue is it’s not understanding “physics” at all! Most of these structures are generated using X-ray crystallography, which is the protein structure in a dry environment with stabilizing salts (so in crystal form), so it’s not even generally representative of the proteins in water!
English
1
0
6
122
Pavle Jeremić
Pavle Jeremić@SynBioMars·
@tszzl @leothecurious Yup! Theoretically you could generate more structural data but the existing structural datasets are biased towards “common” folds in everyday proteins, so by its very nature not extensible to other distributions of structure
English
1
0
6
85
roon
roon@tszzl·
@SynBioMars @leothecurious yeah it’s not a general solve at all, highly biased towards things that *are* tractable and therefore exist in the PDB
English
1
0
3
424