Sam Rodriques @SGRodriques Twitter profile | Pikagi

Pikagi

Sam Rodriques

@SGRodriques

6,705

Followers

368

Following

36

Media

486

Statuses

Director and CEO at FutureHouse. Building an AI scientist.

https://t.co/FIqT8ZCZAe

Joined November 2014

Don't wanna be here? Send us removal request.

Pinned Tweet

@SGRodriques

Sam Rodriques

1 month

Introducing PaperQA2, the first AI agent that conducts entire scientific literature reviews on its own. PaperQA2 is also the first agent to beat PhD and Postdoc-level biology researchers on multiple literature research tasks, as measured both by accuracy on objective benchmarks

78

770

3K

Last Seen Profiles

@littlegoodsheep

@stwmaniax

@bokeplokalmalam

@sakadai_naka

@flingpoly2002

@svetafed12

@AlexisIsOverIt

@mini5gorisuke

@GratisVoel

@stwmaniax

@OCTV2BU

@setwun

@Theworldwuw

@dipperthemenace

@estewe_ibu2

@whitelieorlie

@IszabelMartinez

@bokeplokalmalam

@KxmiKanSama

@stwmaniax

@godinhistory

@na29nay

@mitchnew17

@yujiglow

@grant_mealey

@ACole_World

@yuyan__07

@iamjony94

@luyunxiu175258

@LauraT54761

@ludfernandxs

@staygilmore

@venu16965111

@luyunxiu175258

@mikarika_hiro

@tekkaus

@SGRodriques

Sam Rodriques

11 months

Somehow, the boomer/doomer divide over nuclear meant that we ended up *with* nuclear weapons and *without* (for the most part) nuclear power, which is like the most insane outcome. We should not make the same mistake for AI.

20

211

1K

@SGRodriques

Sam Rodriques

1 year

Today, we are announcing Future House, a philanthropically-funded moonshot focused on building an AI Scientist. At Future House, our 10-year mission is to build semi-autonomous AIs for scientific research, to accelerate the pace of discovery and to provide world-wide access to

64

155

796

@SGRodriques

Sam Rodriques

4 years

New paper out today in @NatureBiotech with Linlin Chen, Fei Chen, @eboyden3 , and many other wonderful coauthors, demonstrating how we can encode the age of an RNA into its sequence to achieve "temporal transcriptomics"

11

158

549

@SGRodriques

Sam Rodriques

10 months

Today, Future House is announcing WikiCrow, our first automated system for synthesizing scientific knowledge. As a demo, we are releasing a cited technical summary for every gene in the human genome that previously lacked a Wikipedia article: 15,616 in all. As scientists, we

14

85

437

@SGRodriques

Sam Rodriques

2 years

Some thoughts on why progress in biology is so slow. Suppose we *actually* want to cure all diseases by 2100, what should we do? We need to be able to do experiments faster in human biology, and we need a better approach to knowledge and talent.

Tweet card media

Why is progress in biology so slow?

I have spent a lot of time recently thinking about why biology is hard. The goal of biomedical research is nominally to cure disease. However, most people don’t seem to be very serious about it. When...

www.sam-rodriques.com

21

61

339

@SGRodriques

Sam Rodriques

3 months

Today, we're releasing LAB-Bench, a set of >2000 evaluations for language models and agents on scientific research tasks in biology. Public models underperform PhD/postdoc-level humans on nearly all tasks. Claude 3.5 Sonnet is the clear frontrunner atm, but long way to go. 1/

Tweet media one

7

67

315

@SGRodriques

Sam Rodriques

1 month

We spent a lot of effort on making our open source version be excellent. We put together a new system of building metadata of arbitrary PDFs, full-text search, and more. See it here: 4/

Tweet card media

GitHub - Future-House/paper-qa: High accuracy RAG for answering questions from scientific documents...

High accuracy RAG for answering questions from scientific documents with citations - Future-House/paper-qa

7

33

239

@SGRodriques

Sam Rodriques

3 years

My first publication with my new affiliation! Check out our comment piece on the FRO model in @Nature with @AdamMarblestone @AGamick @tkalil2050 @C9Martin @MWCvitkovic

6

31

212

@SGRodriques

Sam Rodriques

9 months

At FutureHouse, our AI Scientists need to be able to plan complex biological experiments and analyses, but frontier language models are not very reliable for planning long biology protocols. To that end, we recently created BioPlanner, an automated approach for assessing and

Tweet media one

2

31

199

@SGRodriques

Sam Rodriques

8 months

The default approach to solving biology with ML is to gather as much data as possible as fast as possible and train a Really Big Model. The problem is: if you want to cure diseases, extend lifespan, and engineer human biology, the dataset you want is millions or tens of millions

21

19

198

@SGRodriques

Sam Rodriques

2 years

Priorities for biology are set by people who have direct experience of disease; but research is done by young people, who don't. If we want to keep brilliant people coming, we need to motivate them with bigger visions. A catalog of big visions for biology:

Tweet media one

2

40

164

@SGRodriques

Sam Rodriques

2 years

We desperately need a talent agency for elite STEM talent. Identify extraordinary grad students/postdocs, place them with ultra-promising startups and labs. Probably enough business here for a healthy for-profit, and I know a lot of people who would be customers/invest. 1/2

30

9

161

@SGRodriques

Sam Rodriques

5 months

As a PI in academia I was used to being the person in the room who knew everything. At futurehouse everyone we hire seems to know 10x about basically every topic. It’s wild, still getting used to it. Still can’t believe I get to work with such great people every day.

7

2

154

@SGRodriques

Sam Rodriques

1 month

PaperQA2 finds and summarizes relevant literature, refines its search parameters based on what it finds, and provides cited, factually grounded answers that are more accurate on average than answers provided by PhD and postdoc-level biologists. When applied to answer highly

10

19

154

@SGRodriques

Sam Rodriques

2 months

Several people have asked me to respond to Sakana's AI Scientist post, but judging from comments here and elsewhere, I think people already get it: Sakana’s AI Scientist is just ChatGPT (or Claude) writing short scripts, making plots, and grading its own work. It’s a nice demo,

Tweet media one

9

27

154

@SGRodriques

Sam Rodriques

1 year

Recently, the Wellcome Trust estimated it would cost roughly $10B over 17 years to reconstruct a whole mouse brain, due in large part to the cost of proof-reading. My lab has been working with @E11BIO on a new technology that could, in principle, reduce this figure by 1000x. 1/n

3

22

144

@SGRodriques

Sam Rodriques

8 months

When people ask me what FutureHouse is doing, I tell them, FutureHouse is automating discovery. Scaling up biology research. Building AGI for science. Changing the way science is done. And they say, "Okay, but automating discovery for what?" We then sometimes get lost in this

11

8

144

@SGRodriques

Sam Rodriques

4 years

Proud to announce that I'll be founding the Applied Biotechnology Laboratory at the Francis Crick Institute this January. We have openings for Postdocs and PhD students, working in connectomics, vectors for gene therapy, and diagnostics. @TheCrick @CR_UK

10

32

139

@SGRodriques

Sam Rodriques

1 month

Also, see our preprint for details, here: 5/

Tweet media one

3

13

137

@SGRodriques

Sam Rodriques

3 years

It’s amazing to see the FRO concept get off the ground. Congratulations to @AdamMarblestone , @AGamick , @SchmidtFutures , and everyone else involved!! Everyone needs to be paying attention to this. For people who aren’t familiar with FROs, I’ll provide some background here.

2

36

127

@SGRodriques

Sam Rodriques

1 month

And, of course, this was all made possible through the wonderful generosity of Eric and Wendy Schmidt, and all of our other funders, including Open Philanthropy for supporting our work on LitQA2, the NSF National AI Resource program, and others! 6/

6

2

125

@SGRodriques

Sam Rodriques

10 months

New experiments in the way we do research, from Focused Research Organizations and for-profit institutes to new ARPAs, are changing the landscape of scientific research today. I discuss these trends, the background, and future prospects in my new commentary in Cell, out today:

2

29

115

@SGRodriques

Sam Rodriques

11 months

Today, we are opening recruitment for our Assessment Team at Future House. We want Future House to be the world leader for evaluating the scientific abilities of AI systems, particularly for biology research. Our Assessment Team will make that happen.

4

17

108

@SGRodriques

Sam Rodriques

1 month

PaperQA2 can also do broad-based literature reviews. WikiCrow, which is an agent based on PaperQA2, writes Wikipedia-style articles that are significantly more accurate on average than actual human-written articles on Wikipedia, as judged by PhD and postdoc-level biologists. We

Tweet media one

2

10

107

@SGRodriques

Sam Rodriques

5 months

Key thing to remember: biological intelligence correlates primarily with neurons per unit mass, not total neurons. (Most neurons probably engaged in sensory/motor.) Crow has 1.5B neurons and weighs 1lb. Human has 80B and weighs 150lb. 🐦‍⬛/acc

@natfriedman

Nat Friedman

5 months

Crows have about 1.5B neurons. Assuming each neuron connects to ~7k other neurons on average, that's roughly equivalent to 10T parameters in an artificial neural network. Is this a useful analogy?

61

26

583

13

11

103

@SGRodriques

Sam Rodriques

6 months

I think the reason automating science is hard is that protocols often have 20+ steps; but each individual step will fail or require modification in 5% or 10% of cases. So the probability you get all the way through without having to modify in some way is basically 0. E.g. in

3

11

99

@SGRodriques

Sam Rodriques

11 months

It's no secret that I struggled with innovation & spinout culture in the UK. However, the UK has two major advantages: #1 is that policy reviews often propose bold action, and govt often adopts them. The recommendations here are spot on:

Tweet card media

Independent review of university spin-out companies

An independent report by Professor Irene Tracey and Doctor Andrew Williamson of university spin-out companies and the government’s response.

4

14

94

@SGRodriques

Sam Rodriques

3 years

I think my lab would be best described as an Unfocused Research Organization.

2

1

94

@SGRodriques

Sam Rodriques

3 months

LAB-Bench is, to my knowledge, the first set of evals designed to measure whether models/agents can do scientific research, not just whether they know assorted trivia about science. Procedural evaluations like this for complex tasks will be very important going forward.

@SGRodriques

Sam Rodriques

3 months

Today, we're releasing LAB-Bench, a set of >2000 evaluations for language models and agents on scientific research tasks in biology. Public models underperform PhD/postdoc-level humans on nearly all tasks. Claude 3.5 Sonnet is the clear frontrunner atm, but long way to go. 1/

Tweet media one

7

67

315

2

10

91

@SGRodriques

Sam Rodriques

6 months

I always ask people who the most creative young researchers are in AI for science today. In chemistry, one name has come up no fewer than 7 times: Ryan-Rhys Griffiths, who did chem design w/VAEs, benchmarks for math, and more. Welcome @Ryan__Rhys , who joined FutureHouse today.

3

2

87

@SGRodriques

Sam Rodriques

1 year

The Focused Research Organizations that @AdamMarblestone , @AGamick , and others at @Convergent_FROs have been spinning out have been a huge hit. What have we learned from them, and what should we be doing differently? I think there are three lessons: 1/5

1

22

84

@SGRodriques

Sam Rodriques

1 year

I often hear proposals to build new foundation models for biology. Here are the questions I ask. 1/n

Questions to ask about building foundation models

I often hear proposals to build new foundation models for biology. Here is the list of questions I ask. I rarely get past question 1.1. What is the core task that the model performs/the core thing...

www.sam-rodriques.com

2

5

80

@SGRodriques

Sam Rodriques

1 year

I have been a PI for just over 2 years. My biggest surprise -- which shouldn't really be a surprise in hindsight -- is that academia is really fundamentally an educational institution first, and a research institution second.

4

6

77

@SGRodriques

Sam Rodriques

3 months

We're expanding our wet lab at FutureHouse, and looking for exceptional junior bio researchers. Our AI systems are designing protocols & experiments to make basic science discoveries in biology. If you want to see the future of wet lab research, apply:

0

25

77

@SGRodriques

Sam Rodriques

2 years

My lab is hiring! We're located at the Crick Institute in Central London -- if you want to invent new biotechnologies to measure and control biological systems, get in touch!

Tweet media one

1

33

76

@SGRodriques

Sam Rodriques

1 year

A ton of amazing people have helped get Future House to where it is today. Besides @andrewwhite01 and @ericschmidt , @kulesatony has been a spectacular supporter from the beginning. If I were starting a for-profit, I would take his money any day, the value add is gigantic.

4

6

75

@SGRodriques

Sam Rodriques

4 years

Just arrived in the UK, preparing to start the Applied Biotech Lab next week (). Certainly a difficult time to start something new, but excited to be working on a range of technologies that could have an impact on the current crisis...

7

7

74

@SGRodriques

Sam Rodriques

10 months

The concept of RAG agents like WikiCrow is now well-explored, with systems like Elicit, Perplexity, and others. However, WikiCrow stands out for its scale and accuracy. PaperQA, the agent underlying WikiCrow, matches humans on a new, very hard RAG benchmark we developed, and is

2

8

73

@SGRodriques

Sam Rodriques

1 year

Hard to imagine someone more visionary than Andrew to have on board at Head of Science. Super excited to be building together.

@andrewwhite01

Andrew White 🐦‍⬛

1 year

The last few months, I’ve been helping develop the mission for @FutureHouseSF - a non-profit AI Biology institute. Building on our work with autonomous agent driven chemistry and robotics, we’re building an entire org around the mission to build autonomous scientific agents 1/5

Tweet media one

26

73

537

4

4

71

@SGRodriques

Sam Rodriques

3 years

I started a blog! I'll be discussing sometimes-serious and sometimes-whimsical ideas for reorganizing science research and funding, and occasionally I might throw in some actual science. Today's topic: academic labs that outlast their founders.

Blog | Sam Rodriques

www.sam-rodriques.com

1

8

73

@SGRodriques

Sam Rodriques

2 years

It's happening people

0

8

70

@SGRodriques

Sam Rodriques

9 months

What should policymakers do to maximize the benefits of AI for biology, medicine, and healthcare? I wrote up some recommendations over the summer for focus areas, and just put them on my blog. Interested in feedback, let me know what you think. Bottom lines:

4

13

68

@SGRodriques

Sam Rodriques

4 months

biggest culture win at futurehouse so far is that everyone either pipettes or commits

4

1

68

@SGRodriques

Sam Rodriques

5 months

Actually, the fact that (1) taller humans have bigger brains and more neurons, and yet (2) intelligence/memory do not correlate with height, even though height can vary by 2x, seems to be a deep and fundamental mystery about human intelligence that doesn’t transfer from AI (yet).

18

4

67

@SGRodriques

Sam Rodriques

2 years

🚨My lab is hiring a project manager!🚨 This is an experiment in how we can do academic science better and faster. If you have exceptional organizational skills and want to work alongside inventors to accelerate biotechnology, apply now! 1/n

2

24

65

@SGRodriques

Sam Rodriques

3 years

Excited to announce that my lab was awarded an Impetus Grant yesterday for our work on temporally resolved RNA sequencing methods, which we will use to study cellular senescence. We are hiring! If you're excited about aging, omics, or entrepreneurial biotech, get in touch.

4

1

66

@SGRodriques

Sam Rodriques

8 months

At FutureHouse, we always try to find brilliant people before they're famous: Shannon before information theory, or Einstein before relativity. It's the biggest win when we do. If you don't know about her yet, you will soon. @MichaelaThinks signed with FutureHouse last week.

4

4

63

@SGRodriques

Sam Rodriques

11 months

Robert Boyle's scientific to-do list from the 1600s: -A perpetuall light -The making of glass malleable -A ship to saile with all winds -The art of flying Guys we have done so well.

Tweet card media

What scientists want: Robert Boyle’s to-do list | Royal Society

The to-do list written by Robert Boyle FRS for the early Fellows of the Royal Society, from the prolongation of life to 'varnishes perfumable by rubbing'.

royalsociety.org

2

6

64

@SGRodriques

Sam Rodriques

3 years

Today is my lab's one year anniversary!! Super grateful for all the brilliant and risk-taking students, postdocs, and colleagues I've had the chance to work with, and what a great day for our first publication! This has been an amazing start to an incredible adventure.

1

0

61

@SGRodriques

Sam Rodriques

11 months

People ask me often about how to start a FRO. I wrote down some thoughts about the 4 most important components: mission, outputs, culture, and talent.

Some tips on starting a FRO

Lots of people have approached me recently to ask for advice about how to start a Focused Research Organization. Here are the most important things to pay attention to, in my opinion. Note: Future...

www.sam-rodriques.com

1

10

60

@SGRodriques

Sam Rodriques

4 months

turns out there is a big difference between the things you need to do to please investors (short term, flashy, hype hype hype) and the things you need to do to attract talent (longer-term, impressive, ambitious). Talent is usually more discerning

3

2

61

@SGRodriques

Sam Rodriques

3 years

Wow!! This is really amazing. This looks like a single molecule protein sequencer that is actually going to work! (And remarkably similar to , with @AdamMarblestone )

Tweet card media

A theoretical analysis of single molecule protein sequencing via weak binding spectra

We propose and theoretically study an approach to massively parallel single molecule peptide sequencing, based on single molecule measurement of the kinetics of probe binding (Havranek, et al., 2013)...

journals.plos.org

0

11

57

@SGRodriques

Sam Rodriques

10 months

First challenge in the Future House challenge book officially crossed off. On to the next one...

Tweet media one

4

0

55

@SGRodriques

Sam Rodriques

2 years

In addition to hypothesis-driven science, I think we should also recognize, as a distinct category, journal-driven science, i.e., science you do primarily because it is likely to get published in Science, Nature, or Cell.

4

5

57

@SGRodriques

Sam Rodriques

7 months

🫡

@ericschmidt

Eric Schmidt

7 months

I visited @FutureHouseSF this week, and was amazed by the progress @SGRodriques , @andrewwhite01 , and the others are making. They are extremely fast. A new way of doing science!!!

18

14

164

2

0

56

@SGRodriques

Sam Rodriques

4 months

I recently had a conversation with a friend who wants to start a brain computer interface company about what the challenges and opportunities are, and what is possible on the 50-100yr timescale. Sharing them here: 1/n

Tweet media one

3

7

56

@SGRodriques

Sam Rodriques

2 months

FutureHouse is hosting an AI for Science happy hour next Thursday, the 22nd, at 5pm, for people who are doing cool research or building cool things. Will be in San Francisco. Sign up here:

Tweet media one

3

2

56

@SGRodriques

Sam Rodriques

1 month

figuring out if anyone has done something before is always such a pain...

@andrewwhite01

Andrew White 🐦‍⬛

1 month

we made this:

14

38

289

0

8

55

@SGRodriques

Sam Rodriques

1 year

Some more details on the kinds of things we are thinking about, for people who want to know more :-)

@FutureHouseSF

FutureHouse

1 year

And we’re off! Super excited to be launching today. Over the next 10 years, we are aiming to build a fully autonomous AI Scientist, and to use that AI Scientist to scale up biology research. Here are a few things to expect from us: 1/5

Tweet media one

3

34

175

2

3

53

@SGRodriques

Sam Rodriques

1 year

Huge thanks to @ericschmidt for all of his support. Super excited about all of the work we are going to do together.

@ericschmidt

Eric Schmidt

1 year

Incredibly excited about this; will move biology and AI forward in a very new way. Congratulations to Sam and the team !!!

15

40

337

0

1

53

@SGRodriques

Sam Rodriques

7 months

The time to really worry about risks of new AI-designed bioweapons is once AI-designed drugs get really good. If you can design good drugs, you can def design good bioweapons. And, if you don't understand human biology enough to design drugs, probably hard to design bioweapons.

5

1

51

@SGRodriques

Sam Rodriques

11 months

Since @kulesatony is on the Future House board, we benefit from all the resources @pillar_vc offers their founders, and I swear, starting a new org with Pillar is like easy mode.

1

2

51

@SGRodriques

Sam Rodriques

6 months

What are the key problems we need to solve to accelerate the design-build-test cycle in biomedicine today? 1/

Tweet media one

1

8

51

@SGRodriques

Sam Rodriques

3 years

Excited to tease the next big invention coming out of my lab: high-throughput screening of single graduate students in droplets.

1

4

50

@SGRodriques

Sam Rodriques

2 years

I spent a chunk of 2022 thinking about consciousness: whether it is possible to measure it, and if so, how. I have now written up my thoughts in a super raw format here: 1/n

Tweet media one

7

4

50

@SGRodriques

Sam Rodriques

3 years

Very excited to announce that @neuroluci and I have been named 2021 Allen Distinguished Investigators! Looking forward to working with Lucia to develop new connectomics methods and use them to understand how fly larvae jump. We are hiring -- if it sounds cool, get in touch!

@AllenInstitute

Allen Institute

@AllenInstitute

3 years

How does a new kind of behavior arises in evolution? Allen Distinguished Investigators @neuroluci , & @SGRodriques at @TheCrick are studying larvae brains of a species of fruit fly that have evolved to jump 10 times their body length. #FrontierScience 🧵 3

Tweet media one

1

2

19

0

3

47

@SGRodriques

Sam Rodriques

11 months

Great article about Future House in @endpts by @AndrewE_Dunn today, with more detail about what we're doing from a biotech angle.

Tweet card media

Backed by ex-Google CEO Eric Schmidt, a new nonprofit looks to 'solve science' with AI

Much of the hype in biopharma AI is built around changing the world of drug development. A new nonprofit thinks the right place to start is by building a better lab assistant. San Francisco-based...

0

2

47

@SGRodriques

Sam Rodriques

11 months

If @sama taught us anything, it is that Machiavelli missed a trick when he said it is better to be feared than to be loved.

2

5

46

@SGRodriques

Sam Rodriques

10 months

. @tkalil2050 is probably the most underfollowed person on science X per unit impact.

4

3

44

@SGRodriques

Sam Rodriques

1 year

Time for another round of acknowledgments for some of the other incredible supporters who have helped us along this journey! This includes @ericschmidt , of course, whose vision and generosity is enabling us to do the work we're doing today.

2

0

45

@SGRodriques

Sam Rodriques

2 years

The UK gov't commits to exploring new institutional models for science, including FROs. Pretty surreal to see the PM endorse an idea I proposed in my PhD thesis! A testament to the amazing vision and execution of @Convergent_FROs @AdamMarblestone @AGamick

2

4

45

@SGRodriques

Sam Rodriques

2 years

Hot take: You could prevent 80% of the mortality in the US with cures for just 10 diseases. Biotech doesn't need scale; we just need a few really, really good shots on goal.

3

3

43

@SGRodriques

Sam Rodriques

10 months

Coming now from the bio side, WikiCrow is actually pretty awesome for coming up with molecular hypotheses. An experiment in my Crick lab turned up a bunch of genes we had never heard of yesterday, and we used the WikiCrow articles to hypothesize an apoptosis pathway from them...

2

1

42

@SGRodriques

Sam Rodriques

8 months

Things we learned starting FutureHouse #1 : You can hire people who know more than you do*. FutureHouse launched in September. Prior to that, I was an academic. In academia, you are a teacher, and so you are always hiring people who are less accomplished than you are. You hire

0

1

41

@SGRodriques

Sam Rodriques

4 months

I often talk to people who think biotech and synbio should be more ambitious, dream bigger. Using biology to terraform mars -- sure fits the bill.

@erika_alden_d

Erika Alden DeBenedictis

4 months

Introducing Pioneer Labs, a startup engineering microbes for Mars 🚀🌼 @Pioneer__Labs

18

34

223

2

0

42

@SGRodriques

Sam Rodriques

2 years

I previously posted a set of real-world science questions to evaluate ChatGPT. The GPT-4 results are now in: it is slightly better, but lots of room for improvement. SCORES (Incorrect / Partial / Correct / Too hard to grade): ChatGPT Original: 18 / 4 / 2

Tweet media one

7

13

42

@SGRodriques

Sam Rodriques

11 months

In the run up to Thanksgiving, I want to take a moment to emphasize my deep appreciation for @richardfuisz , who played an integral role in getting Future House off the ground. 1/n

1

1

41

@SGRodriques

Sam Rodriques

9 months

We need an Addgene for bio datasets

11

4

40

@SGRodriques

Sam Rodriques

3 years

Super excited to say that my lab has recently received multiple large grants, to be announced in the coming months, so we're hiring! If you're looking for a postdoc position, or starting to think about one, get in touch!

0

6

40

@SGRodriques

Sam Rodriques

2 years

Can we build an AI scientist? Existing LLMs are very far away, but progress is fast and it seems possible. We need to figure out how to evaluate new models as they emerge. Thoughts here with many examples. If you're excited about this, get in touch.

Tweet media one

0

8

40

@SGRodriques

Sam Rodriques

11 months

This was not on my 2023 bingo card.

3

1

39

@SGRodriques

Sam Rodriques

3 months

Blog post: PDF: HF: Key results: -Models perform badly on the tool-use benchmarks, because we don’t provide them with the relevant tools. Infrastructure needs to be built. No surprise there. -More

Tweet card media

futurehouse/lab-bench · Datasets at Hugging Face

1

3

38

@SGRodriques

Sam Rodriques

2 years

People are less good at coming up with new ideas over zoom. Cool study lending scientific credibility to something everyone already knew.

2

8

38

@SGRodriques

Sam Rodriques

5 months

Based on Tom & co's trackrecord, high chance this becomes the most important org in US science philanthropy today.

@tkalil2050

Thomas Kalil

5 months

I am excited to announce the launch of Renaissance Philanthropy, with a phenomenal team including @KumarAGarg . Our goal is to help foundations and philanthropists promote a 21st century renaissance, fueled by advances in science, technology and innovation.

22

45

297

0

5

37

@SGRodriques

Sam Rodriques

1 month

The curse of usefulness New ideas that are visionary but useless get evaluated on how visionary they are. New ideas that are visionary and useful get evaluated on how they compare to SOTA. As a result some very visionary ideas that are close to but not yet at SOTA get panned.

0

0

37

@SGRodriques

Sam Rodriques

9 months

1. We need integrated research environments, where biology researchers, AI researchers, and AI engineers can work together. (These need to be equal collabs also, and avoid the "AI researcher as code monkey" or "bio researcher as data monkey" pitfalls). @FutureHouseSF

1

3

35

@SGRodriques

Sam Rodriques

3 months

The team that pulled together LAB-Bench, led by @jonmlaurent and @joejanizek , is extremely cracked. They pulled this together, from beginning to end, in just over 2 months. Lots more to do. If you want to join: Major contributions also from Michael

1

6

36

@SGRodriques

Sam Rodriques

9 months

2. We need the scientific literature to be publicly available. The government paid for almost all of it. We can't get value out of it if we don't have access to it.

1

3

35

@SGRodriques

Sam Rodriques

3 months

Spoke last night to some brilliant undergrads @NucleateDojo in SF about major challenges that UK + Europe face for leading in science and technology. Wake up this morning to find the UK has shelved its AI initiative. Case in point:

Tweet card media

Government shelves £1.3bn UK tech and AI plans

The future of a number of major products - including a new UK supercomputer - are now in doubt.

2

7

36

@SGRodriques

Sam Rodriques

2 years

Has anyone ever tried to put selective pressure on animals to live longer, and see what changes? E.g. only allow animals to mate past prime reproductive age? @LauraDeming @bettslacroix ?

6

2

34

@SGRodriques

Sam Rodriques

1 year

These are probably the two most important XKCD comics to remember these days...

Tweet media one

Tweet media two

1

2

35

@SGRodriques

Sam Rodriques

2 years

The Crick is a spectacular place to start an academic career. Most generous startup resources in the world, outstanding (and very new) facilities, no obligation to teach. Everyone who wants to start an academic bio lab should apply.

@TheCrick

The Crick

2 years

Come start your group at the Crick! 💡 We're looking for creative researchers from all areas of biology and biomedicine to join our early career faculty community. #AcademicJobs | #ECRChat Learn more and apply by 6 October ⤵️

3

174

234

0

8

34

@SGRodriques

Sam Rodriques

2 years

Yesterday, it was -110F on top of Mt Washington (with wind chill). The sublimation point of dry ice is -109F. This made me think: can we remove CO2 from the air by freezing it out? 1/

9

0

34

@SGRodriques

Sam Rodriques

2 years

One of the weird things about the UK is the fact that its universities are super inward-looking: they only tend to compare themselves to other UK institutions, and like to come up with reasons for why it's not fair to compare themselves to their international competitors.

@nathanbenaich

Nathan Benaich

2 years

The UK Research and Innovation, a public body that awards amazing research grants, recently asked: "Is the UK really not as good as the US at turning university research into new businesses?" They concluded: "London is better than the US" Here's why this is misleading:

Tweet media one

11

42

166

2

8

31

@SGRodriques

Sam Rodriques

3 months

We have 8 categories of tasks, all MCQs. FigQA, TableQA, and ProtocolQA are reasoning tasks that don’t require tools. LitQA, SeqQA, dbQA, and suppQA are tool-use benchmarks for literature search, database access, etc. Cloning Scenarios are non-trivial “real-world” challenges. 2/

Tweet media one

2

1

33

@SGRodriques

Sam Rodriques

3 months

FutureHouse is hosting a happy hour on Wednesday at ICML. Come find us...

Tweet card media

FutureHouse ICML Happy Hour | Partiful

3

3

32

@SGRodriques

Sam Rodriques

6 years

Unlike DNA, there's no way to copy proteins, so protein sequencing is intrinsically a single molecule problem. But it might be more possible than one would otherwise assume, as @AdamMarblestone , @eboyden3 and I have calculated.

@eboyden3

Ed Boyden

6 years

Next gen DNA sequencing has revolutionized biology. But what if you could sequence individual proteins? Here we propose, and mathematically evaluate, a concept of how to do that.

4

83

218

2

16

30

@SGRodriques

Sam Rodriques

1 year

E11 has the best shot right now at mapping the whole brain. This is the NGS of connectomics. I would definitely jump on this team if I could.

@junaxup

Jun Axup

1 year

✨👩‍🔬 Do you love organizing, optimizing processes, and helping to enable impactful science? @E11BIO is looking for a lab operations manager to join our team! #hiring

Tweet media one

0

10

12

0

9

30