First, I’m now convinced that ChatGPT understands what it reads. Second, reasoning models persuade me that ChatGPT is creative. Third, ChatGPT summarizes texts extremely well, which I believe to be a robust measure of intelligence.

At the same time, I don’t believe in “general intelligence” (I don’t consider myself one) and so I don’t believe the idea of “AGI” is very meaningful.

Finally, AI products now pay for AGI research, which means that AI has reached early stages of the long-awaited self-improvement loop. I believe that this makes a lot of the discussion of “AGI” and “superintelligence” timelines obsolete.

I expand on all of these points below.

ChatGPT understands what it reads

For me, the question of whether we’re making progress towards AGI has always been about understanding. Does ChatGPT really understand or is it just a Chinese room stupidly mapping inputs to outputs? I now think it does understand.

What really convinced me was this one tweet. In it, François Fleuret makes fun of the OpenAI o1 model for failing to notice that the logical puzzle he asked it to solve had a trivial solution.

I remember first seeing this tweet in September 2024 and laughing at it too, thinking what a perfect demonstration of how dumb LLMs actually are.

In April 2025, I stumbled on that tweet again and thought, why not see if OpenAI o3 would make the same mistake? And it did. o3, a model that is described by one of the greatest living mathematicians as being on par with a not-completely-incompetent mathematics graduate student, cannot solve a literally trivial puzzle.

But then I started thinking. What if the model did understand the puzzle and just didn’t pay enough attention to it? So I asked it to read the question more carefully. And guess what. It solved it. As soon as o3 read carefully, it had no problem solving a novel puzzle it’s never encountered before.

Another example. These questions by Quintus. They’re his joking attempt to construct a series of the Law School Admission Test-style questions based on my “I’m broke, divorced, and unemployed” piece. These are very difficult questions about the meaning of a text ChatGPT 100% never saw before. It answers most of them right. I simply don’t see how it would be possible without it possessing some kind of understanding of the text.

Another example. I was writing a short piece recently. I asked a very smart friend of mine what they thought the main point of the piece was. They got it wrong. I asked ChatGPT what the main point of the piece was. ChatGPT got it right. Again, the question provably was not answered anywhere ever & was fully self-contained. The only way to learn the main point of it would be to read it and, I believe, to understand it. My very smart friend failed. ChatGPT succeeded. If this doesn’t convince you that ChatGPT can generalize outside its training distribution, I don’t know what would.

Or take this greentext from GPT-3:

Can you read this greentext and say that ChatGPT is a stochastic parrot with no sense of what it says? I can’t.

For me, these examples make an incredibly convincing case that ChatGPT is an actual intelligent entity and that therefore we are well on the path to building AGI.

Interlude: women are good for you

A reasonable question to ask here would be: How come I needed to be convinced that ChatGPT understands today, 3 years later?

Large language models exhibited all of these signs of understanding years ago! The greentext I just presented, for example, is from 2022. I spent a ton of time with LLMs back then and used GPT-3 to help me write multiple blog posts in 2022 and 2023.

I think what happened is that I got psychologically overwhelmed by my previous apocalyptic-style belief in AGI and getting fired from OpenAI in late 2023. I guess it was basically a nervous breakdown.

The combination of thinking that AI is going to radically change (or destroy) the world in 3-5 years combined with me not being able to do anything about it (on account of getting fired) paralyzed me and at some point my brain just blocked any thoughts about AI.

As a result, for most of 2024, I basically stopped using ChatGPT and the only news about AI I would pay attention to was how it’s not happening.

Around October 2024, I started to get seriously worried that the reason I’m not thinking about AI is for the reasons I just noted. I figured I should get a job at OpenAI again to make sure I don’t have any residual trauma associated with it & I did it in January 2025.

I also started dating a beautiful girl completely outside of the AI bubble but who uses ChatGPT for about 12 hours a day and couldn’t help but get curious about this. Looking at her ChatGPT usage, I just couldn’t keep maintaining that belief that AI is not happening.

The combo of OpenAI + my girlfriend got me emotionally unstuck and I started to be able to notice all of these pieces of evidence that I wasn’t considering before.

It is also remarkable to me just how differently I interpret the same evidence today compared to 6-12 months ago.

The logical puzzle I mention in the beginning of this post is one example. I initially collected it as an example of how stupid ChatGPT was. Then I looked at it 8 months later and it became an example of how smart ChatGPT is.

Or take the famous “9.11 or 9.9 what’s bigger” question.

I used to laugh at ChatGPT every time I saw it. But I thought about it more recently and concluded it’s a problem of context, not intelligence. There truly are many instances where 9.11 is bigger than 9.9 (books, academic papers, software versions). It’s not unreasonable for ChatGPT to think that the question “9.11 or 9.9 what’s bigger” is about a context in which 9.11 is bigger than 9.9, if no other information is given to it.

Relatedly, in 2024 I collected a bunch of examples of ChatGPT failing on seemingly simple tasks. Most of them are now solved, usually one-shotted by ChatGPT. Example 1. Example 2. Example 3.

It’s become very difficult for me to maintain the belief in the stupidity of ChatGPT when every time I laugh at it, it ends up ridiculing me 6 months later.

Large language models are creative

or: LLMs + RL = intelligence move 37

Okay, so what if large language models like ChatGPT understand? Maybe they still can’t come up with new stuff and do science. Dwarkesh famously asked “what should we make of the fact that despite having basically every known fact about the world memorized, these models haven’t, as far as I know, made a single new discovery?”

RL working on LLMs – pioneered by OpenAI’s reasoning models – convinced me that this is not a problem because whenever RL works it discovers new and creative solutions. Chess (do watch the video). Go. Math. Physics simulations. Video Games. Anything.

RL working = creativity. This is The Paradigm.

For me, it all came together when I read Andrej Karpathy’s twitter essay on the move 37:

“Move 37” is the word-of-day - it’s when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only achievable by large-scale reinforcement learning. You can’t get there by expert imitation. It’s when AlphaGo played move 37 in Game 2 against Lee Sedol, a weird move that was estimated to only have 1 in 10,000 chance to be played by a human, but one that was creative and brilliant in retrospect, leading to a win in that game.

We’ve seen Move 37 in a closed, game-like environment like Go, but with the latest crop of “thinking” LLM models (e.g. OpenAI-o1, DeepSeek-R1, Gemini 2.0 Flash Thinking), we are seeing the first very early glimmers of things like it in open world domains. The models discover, in the process of trying to solve many diverse math/code/etc. problems, strategies that resemble the internal monologue of humans, which are very hard (/impossible) to directly program into the models. I call these “cognitive strategies” - things like approaching a problem from different angles, trying out different ideas, finding analogies, backtracking, re-examining, etc.

Weird as it sounds, it’s plausible that LLMs can discover better ways of thinking, of solving problems, of connecting ideas across disciplines, and do so in a way we will find surprising, puzzling, but creative and brilliant in retrospect.

As X. Dong from NVIDIA notes about a recent result, “The RL-ed model makes great progress on some problems that the base model has no understanding of, regardless of how many times it tries.”

All of this creativity discussion also reminds me of Tyler Cowen’s point that he got AI-pilled decades ago when he saw AI beating humans at chess. People thought that chess was incredibly intuitive and so once Tyler saw AI beating us at chess he figured it’ll beat us at other intuitive things as well (1, 2, 3).

Compression is intelligence and ChatGPT compresses really well

When I want to figure out how smart someone is, I often ask them: “what’s the main point of X (e.g. an essay)?”. What the question really asks is to compress many words (an essay) into few words (a sentence), while discarding everything unimportant. If the person I’m talking to can do it, I conclude they’re smart. If they can’t, the reason I’m still talking to them is probably because they’re very good-looking. (this is also why I much prefer short books – I suspect they’re written by smarter people)

So when I ask ChatGPT what the main point of an essay is (as I did in one of the examples here earlier) and it succeeds when my very smart friend fails, I conclude that it is really good at compression and is really smart.

(There’s a very closely related point about prediction ability being an indicator of intelligence. I think compression and prediction are two sides of the same coin and so it makes sense that ChatGPT learned how to compress the text very well by learning to predict the next word it’s going to see during its training.)

Why I think the idea of AGI is stupid

Because I don’t believe in “general intelligence”. AGI is usually defined in some reference to humans and, for one, I think I am an extremely narrow intelligence.

For example, I can’t multiply 3289 by 5721 by myself. In fact, I can’t even multiply 328 by 572! I can just barely multiply 32 by 57 and I’ll still fail like 20% of the time.

I’ve set my alarm to 6am hundreds of times and I still haven’t managed to figure out that I’m just going to turn it off and keep sleeping until 8am.

I can figure out what my girlfriend is upset about maybe 10% of the time, but most of the time I’m as good at figuring that out as an abacus.

Am I generally intelligent? To me, the answer is obviously no. I can learn how to do a few things here and there. I can learn how to do a few more things with a piece of paper. I can learn how to do a bunch more things by imitating other people or using my computer. But that’s it!

Human civilization and our technological progress were not enabled by any kind of general intelligence! They were enabled by us being able to learn how to do a few things here and there.

It doesn’t matter if AGI is real or not

(Funnily enough, I got this point from a recent Sam Altman blog post.)

For many decades – and as far as a few years ago – dreams and visions were required to secure funding for AGI research. This is why the AI industry used to have periodic winters; this is why no AGI company survived for more than a few years; this is why AGI used to be a dirty word associated with, as DeepMind’s co-founder says, “the lunatic fringe”.

Today, for the first time ever, economic value created as a result of AGI research is enough to support further AGI research. Hundreds of millions of people use ChatGPT every day and tens of millions of people pay for ChatGPT.

I think it’s instructive just how different this self-improvement loop is from the one usually feared. The reason AI products pay for AGI research is because they’re useful to us. Without humans, AI is incapable of producing economic value, because the very notion of economic value is about what humans find valuable. So the ability of AI to improve itself is completely dependent on it continuing to be useful to us.

AI progress is much smoother than anyone would’ve expected and its usefulness depends much more on non-intelligent scaffolding than anyone would’ve expected. So I think at this point it really doesn’t matter when we build “AGI” or “superintelligence” or whatever. AI research pays for itself = AI is here to stay & keep improving.

I always think about technologies of the past that really changed the world. For example, the printing press completely reorganized the world in the 16th-19th centuries, resulting in Christian Reformation, the Hundred Years’ War, modern nation states, and eventual scientific and industrial revolutions. There was 0 intelligence contained in that technology. But it changed the way information travelled, radically increased the power of ideas and the value of literacy, and gave us awesome new powers to wield in pursuit of our goals.

AI is already making both individuals and groups of humans dramatically more leveraged in our ability to affect the world. How we’re going to use that leverage is up to us.

Acknowledgements

Thanks to Sundari Sheldon, Quentin Wach, Victor Li, Emil Ryd, and Xavi Costafreda-Fu for reading drafts and for suggestions.

Introduction

Some topics here I’ve been thinking about for years, some for months, and some for weeks. Some are personal problems, some relate to projects I want to pursue, and some are more philosophical. What unites all of these questions is that, despite my best efforts, I haven’t been able either to answer them or get them out of my head.

If you have thoughts on any of these questions, let me know.

How to fly to the stars?

I want to host a party at the center of the universe. There are several problems with this idea.

First, I’m basically scientifically illiterate and physicists I talk to tell me that the universe doesn’t even have a center. They say that the universe looks the same in every direction from any point regardless of where you start looking from and that therefore the idea of a “center” doesn’t even make sense. This seems like a big problem.

Second, even if we found the center of the universe, it would probably be way too far away from us to get there. The universe has existed for more than 13 billion years. Who knows how far away its center is?

If we’re talking about traveling to Mars, it’s very clear how to do that. Just get on Elon’s Starship and fly. If we’re talking about traveling to the nearest star system, it’s pretty clear how to do that as well. Build a giant sail. Construct solar-powered laser plant somewhere in the solar system. Point the laser at the sail. Get to 90% of the speed of light. Get to Alpha Centauri in less than 5 years. Or if you don’t want to do that, you can blow up a bunch of nuclear bombs and use that to propel your spacecraft.

But even getting to the center of the Milky Way galaxy is another matter entirely. Alpha Centauri is 4 light-years away. The center of the Milky Way galaxy is 26,000 light years away. The only way to get there would be either via (1) a multigenerational spaceship, a faster-than-light spaceship, or (3) by figuring out how to make human cryonics/hibernation work.

As far as I know, we have no idea how to do any of these. And this is just traveling within our galaxy. So I really have little idea how to host that party. We would probably need to figure out some kind of faster-than-light travel (e.g. via wormholes or yet undiscovered physics).

In any case, if you’d like to get added to my Partiful invite, let me know.

Relevant pieces of media: Spore (a 2008 video game), Project X (a 2012 movie), Interstellar (a 2014 movie), Soulless (a 2012 Russian movie [trailer, watch with English subtitles]), Skins series 1 (a 2007 TV show [trailer]).

Interlude: How can I get OpenAI to send me to Rome for a few months?

I’m currently a contractor at OpenAI. It’s a fascinating company and I want to get converted to full-time. At the same time, I really want to spend a few months hanging out in Rome and doing my own thing, talking to people, writing, thinking, etc.

I’m very confident that if I do that while employed at OpenAI, it will somehow be good for the company even though I have no idea how. Everything good that ever happened to me initially started as me just trying to do something very strange or interesting for unexplainable reasons!

In any case, I have no idea how I would pitch this idea to OpenAI, and my best guess is that if I really do try to do that, it will in fact severely jeopardize my chances of getting a full-time offer.

How to defeat death?

(my definition of “defeating death” means “getting to constant probability of death every year of life”. This still implies finite expected lifespan rather than immortality)

I’m very confused about this.

First, here are some answers that I don’t think are right: Christian afterlife. Eastern-style eternal rebirth. Atheist technological singularity bringing Heaven (and/or Hell) to Earth.

If we’re being more down-to-earth, the first thing we want to do is solve all diseases, including aging.

Whenever I talk to people about this I can’t escape the feeling that it’s just too early to be working on aging. I have some kind of intuition that while a problem seems like a scientific problem rather than an engineering problem, you shouldn’t work on it too directly or you won’t explore the space of solutions. And I’m pretty sure aging is a scientific problem for us today.

A few years ago, Ryan Flynn (supported by New Science), currently a Professor at Harvard discovered a new type of RNA molecule (glycoRNA).

In 2024, a group of scientists discovered a new type of viroid (a virus but without the capsid shell) which they called Obelisks in the human body. We have no idea what they do.

Just two months ago, in March 2025, Nature published a piece saying a “textbook assumption about the brain’s most abundant receptors needs to be rewritten”, reporting that GluA2-containing AMPA receptors in the brain are often orders of magnitude more permeable to calcium than previously thought.

I just don’t see how we’re going to make interventions on the human body that affect such complicated processes as aging while we still keep discovering new types of molecules and viral particles in the human body. If it was a monocausal disease, sure. You just find a way to destroy whatever the agent of disease is and you don’t need to know how the body works or what’s happening in it aside from that at all.

But for an endogenously-driven multi-causal process, the origin of which we don’t know, the markers of which are all very sus, the feedback loops on which are measured in years? I just don’t see it happening.

I spend a lot of time thinking about the blog post ‘The “it” in AI models is the dataset.’ and I wonder how much progress we can make given how bad our data collection abilities in biology are. I’m not even saying that we have to “understand” everything that happens in our body. But at least we need to be able to get good data on what’s happening.

I like Ed Boyden’s idea that we want to “see everything and control everything” but we seem to be very far from that. I continue to think that one of the best things anyone could do for science is to just give Boyden $1b/year budget for the next 20 years for his biological tool-building and then in 20 years for-profit companies with whatever that effort enables.

Finally, we haven’t solved a single brain disease ever. I would personally say that it seems to be one of the more important organs out there and our inability to intervene on it makes me pessimistic about our knowledge of how to reverse its aging (which seems like a strictly and significantly more difficult task).

I hope to be proven wrong here and I hope that companies like NewLimit, Retro, and Altos succeed, despite everything I wrote above.

Boris Power notes that I should follow my own advice and reframe “defeat death” into a positive goal like “How to live more?”. Part of living more is living longer. But this also suggests things like richer and more meaningful experiences, augmentation, increasing the capacity of our bodies, etc.

For example, think of someone’s life at 40 years old and 80-years-old. In which ways are the lives of a 40-year-old usually better than an 80 year old? Can we now project this gradient backwards even further? What does life look then?

Interlude: What should I do with my life?

I’m very emotionally attached to New Science and to the idea of building new institutions of science. I’ve been wanting to run a research institute since at least 2018.

I believe that science, as a study of God, is sacred, and there’s no sense of sanctity at all left in modern academia. As far as I can tell, most academics simply do not believe in truth. Academia as a broader cultural entity certainly doesn’t. This pisses me off, and this is why I started New Science in the first place: to figure out how to create institutions that actually believe in the pursuit of truth first and foremost.

But I feel very lost now. AI really is getting incredibly powerful. I think I’m AI-pilled after all. If – that’s an if that deserves good questioning – the main constraint on the progress of science is sheer intelligence, perhaps working on scaling intelligence is the best thing one can be doing. This is what I’m doing now by working at OpenAI.

However, if both AI and robotics are going to make most human scientists unemployed anyway, then what’s the point of building new institutions of science? We’re entering the age of Pure Will.

On the other hand, I don’t see for-profit companies doing basic science research and pursuing truth just for the sake of it – and AI progress doesn’t really change that. So there’s a world in which building new institutions of science (perhaps institutions of learning, a la the first universities?) is the right thing to do after all. I hope we’re in this world. Thinking…

Also see: What Should You Do with Your Life? Directions and Advice, On Impact, Why you shouldn’t build your career around existential risk

What’s the fundamental nature of reality? How to study it?

People used to think that Newton’s laws were how reality worked but it turned out that if you push your system hard enough they break down. Today people think that QM and Relativity theory are how reality works. They debate all kinds of philosophical questions about multiverses, collapse, fundamental randomness or lack of it, based on our construction of these theories. I don’t understand why you’d do that.

I think the right way to think about current scientific physical theories is the same as the way we think of Newton’s laws: good approximations for the range of physical systems we can currently access. At some point in the future, we’ll invent better instruments and find the places where e.g. quantum mechanics breaks down and figure out new theories. And then the same thing will happen to those theories, and the theories after them, etc.

Why not think of known physics as a set of constraints on the real generative function of the universe, whatever it is, instead of the descriptions of the ultimate reality?

I often hear people say things like “consciousness is the only real thing and that’s what determines reality”. I mean sure, there’s a way in which everything we observe is just a product of our consciousness. But there’s a hard limit on how far you can push this argument: you can shoot yourself with a gun and you’re done. That’s it. Your conscious experience ends and yet the world – with all of its natural laws – remains.

So there are hard limits to hyperstition and there really are laws of nature, independent of individual consciousness.

How to find these “final” laws rather than just their approximations…?

Relevant media: Brian Skinner, From Eros to Gaia (1992 book).

Why does everyone I know believe that the world is going to end?

There’s something to the idea of Armageddon.

A couple of years ago I really got into my head the idea that the world will end in a few years due to AI singularity/doom and have been having trouble getting rid of it ever since. Plenty of my friends and people I know have never escaped.

I mean, what if it is true? What if the world really is about to end? Then nothing matters. Nothing except fully dedicating yourself to averting the end of course.

I understand early Christians much better now. If I were living then and saw Jesus doing the miracles and then telling me that the world is ending within our lifetimes, I probably would’ve dropped whatever I was doing, donated all of my money to him, and spent the rest of my life spreading the Word in an attempt to save as many souls as I could.

What I find notable is that in overtaking Christianity as the leading thought doctrine, science, and physics specifically, has replaced the Biblical story of creation, but also its story of Armageddon. Christianity said it was the Rapture. Physics says it’s the heat death of the universe.

Every single story of Armageddon ever told so far has been false. I cannot help but to reject the idea of Armageddon – whether Religious, Technological, or Scientific – entirely. And yet every single person I know still believes some version of it.

I mean it literally: every single person I know believes in either the religious Rapture, the heat death of the universe, or AI Utopia/Doom. Why?

Conclusion

In 2024, I didn’t know what to do, so I ran a Twitter poll posing this question to my followers (vox populi, vox dei). They told me to get serious about physics, so I flew into Boston at the beginning of the Spring term 2024 and spent 4 months auditing physics classes at MIT, going to a lot of research seminars, and talking to physicists.

I ended up auditing undergraduate Quantum Physics I, undergraduate Statistical Physics I, graduate Astrophysics I, and graduate Atomic and Optical Physics (I’m not claiming to have understood much in any of these classes; although, at the insistence of Isaak Freeman (a New Science Fellow), I did show up for one of the Quantum Physics I midterms, had it graded, and successfully passed it, which I think shows that I wasn’t 100% just LARPing).

In any case, it was all very fun, I met a lot of great people, at some point a Harvard physics professor offered to write me a recommendation letter for grad school, apparently being impressed with all of the harebrained questions I was asking at the weekly research seminar he was in charge of, but it became pretty clear that I’m not becoming a Physics Professor any time soon.

I spent June in Mexico City with Misha Yagudin, July to October traveling around wherever people had a couch for me to sleep on, went to New York for November, hung out in Taiwan with Misha in December, but eventually figured that OpenAI would be a good place for me to think about the future of science, and started here in January.

For 2025, my #1 goal is to get out of debt. My #2 goal is to learn to drive. My #3 goal is to get a better feeling for what the the future of science looks like.

I want OpenAI to work out and I would love to continue pursuing my curiosity and thinking about science & AI at the company. If it doesn’t, I’ll probably become unemployed for a bit again (and writing more). What’s afterwards? I’m not sure.

Acknowledgements

Thanks to Boris Power, Jackson Veigel, Maksym Sherman, Mehran Jalali, Sundari Sheldon, and for conversations and feedback.

Notes

• I promised myself to journal for 40 days 1 hour in the morning and 1 hour in the evening to figure out what I want to do and how I’m feeling, triggered by a major personal event. I’m 3 weeks in and it’s been a very interesting experience. Just getting curious about my feelings without any expectation of change, etc.

• “Учёные ищут путь к бессмертию, тогда как смерть и всеобщее равенство перед ней - это единственное, что удерживает наше общество от превращения в ад.”

• Adam Majmudar tells me that if I just keep writing about things I’m thinking abstractly about instead of actually committing to something and then executing on it then I’m never going to get there.

Appendix: 2021-02-25 concerns

[this is from my personal notes written in February 2025; I announced New Science publicly on May 13, 2021]

ok i feel terrible like my project doesn’t make any sense at all in the first place…….

ok so what’s my project? building new scientific institutions, decoupled from existing ones

but you know what existing scientific institutions work totally fine

we have a shitton of basic science

the gap is translation

the gap is just people

the gap is not in us not having enough fundamental science or us not moving quickly enough

anyone really smart can get into science or if they don’t want that get a job at a biotech

besides there’s the whole issue of biosafety, maybe it’s more important to worry about it than about speed of science itself

also church lab exists - well but it will soon not exist!

so yeah this doesn’t really make sense to work on

basic science is bs lol

tony kulesa is in biotech

everyone else too

i don’t see a reason for why i would try to build new institutions of basic science

also there’s max planck institute in germany

given how few really brilliant people there are, i think we are doing good

…

yep……

George Church exists

“if you have low gpa you’re fucked” no you’re not, you work as a tech for 2 years and you’re golden

perhaps what i should do is find a goal that’s really helpful?

not just abstarctly build new scientific institutions…..

science is really doing fine

there’s thought emporium

ok but there’rs adam strandberg

what’s up with him? well yeah i can help him i guess?

but as i myself said, he’ll find a way

and he’s a big exception

and yes church is an exception

but what about adam marblestone, he didn’t get in

and in europe, there’s Crick institute!!!! take brilliant people, get them money

plenty of places to work on crazy stuff

shoudl i expose the corruption of NIH? that’s how everyone makes a name for themselves…!

• Adam Strandberg did not get NSF

there’s literally just one Boyden and just one Church

i refused an offer from church 1.5 years - he was interested in me being a fundraiser for him. this would actually be perfect i thought he was too old and too rich i should’ve thought more about this now i feel terrible lol

wait but biotech is not working on basic science lmao or Boyden or Open Phil… I firmly believe they could’ve brought me in and it would’ve been great for them.

ok but what about the publication system

you need to publish in CNS in order to graduate and do cool shit

anyway eLife is doing great there I think and the publication system becomes better too

but anyway what does new institutions of basic science even mean

suppose i have $10b per year budget

what exactly will be better in my system? or do i really believe competition is all we need?

more people should be going into biotech anyway?

are there many people stuck in tech positions? well i’m not aware of them!

like i’m just making an argument that it’s not enough

also jed was right – i need to figure out a way to justify my project. and if i couldn’t justify it then probably i just didn’t have a very good idea of what i’m trying to do!

like how do you even improve academia?

if we don’t even know what the problems are, how am i saying i’m going to build new academic institutions

if career progression is to be decoupled, it needs to be a separate system!!!

you can always get a job at a biotech these days anyway!

MOTIVATION PABLO ADAM STRANDBERG JP BIDA ADAM MARBLESTONE CHURCH Andrew York?

at the end of the day, everyone i talk to went to a top school! can’t turn this around!!!

like who are tehse mythical people who i want to support??? i can’t find them!!!!

but maybe academia is good – motivates people to work really hard!!!!!

idk man lol

maybe i can do this new science research and shit and then start a biotech fund lol –

ok let me find the most motivational aspects for me!! —–

the other claim is that i need to enable just a very small number of individuals

call with egan? look at it

look at the original essay version

also what about the whole issue of discoveries following one another…….

https://docs.google.com/document/d/1vk4eKUbW9K6jg1QpQjeUbbDoQga6TqD1wVh9FwOMJ1U/edit#heading=h.27l98jja95yi

Noah reminded me that there’s so much fucked up shit with academia….that’s incredible. like grant cut offs… again, the entire competition system of academia… it just makes so little sense….it’s incredible… also I’m talking to people working in hot areas…but in not hot areas…shit’s fucked up…nobody is researching cryonics for example!!! ok, I’m inspired now. lol

the reason for my academia:

• it’s just going to be better

• the young dude from oxford - in seoul

• those who are not politically savvy are fucked!!!!!

2021-02-27 i feel terrible anxiety for not delivering anything with new science in so long…

i just need to find one crispt, to enable a few lost geniuses

https://docs.google.com/document/d/1CcOR4HERqVcRiZU6Nc37VckTB5Ywa-QWPNdYC2EFOso/edit#

• get more people working on biology because we need immortality, self-replicating bots etc!!!!

• the existing academia is just not big enough - that’s enough justification

• we need more people working on bio!!!!!! get everyone from google to work on biology lmao

expanding into space, colonizing mars is INCREDIBLY INSPIRATIONAL

• ok we need to live forever

• we don’t yet know the answers

• we need basic science: see crispr is a perfect example - defense mechanisms in bacteria used for editing genes - completely unexpected maybe i should do a PhD in biology lol

The absolute pillars:

• I want scientists to be able to become PIs at 25

The pillars:

• I want scientists to be able to project their budgets into the future

• I want scientists to pick their grad students

• colonize space is

• important things:

  • artificial organs

  • regeneration

  • get more people working in research

  • become rich to get the ability to new technologies first!!!

  • prepare for AGI

  • gradually replace aging cells with good cells

Appendix: how to speed up the future?

See here (due to Substack email cutoff: https://guzey.com/what-im-thinking-about/#appendix-how-to-speed-up-the-future)

Every time I talk to a biologist they say "yes, the NIH is horrible, it's impossible to get anything interesting funded by it, it's getting worse by the year, and God I hope Congress increases its budget because my career depends on it."

The NIH is a tyrannical, capricious, self-serving $50 billion a year Kafkaesque Leviathan ruled over by a clique of septuagenarians who couldn't care less for science or for scientists.

The reason biomedical science advances one death at a time is because if you're not about to die the NIH will not give any money. It provides 7 times more funding to scientists 66 years old and older (literally retirement age) than to those 35 and younger. And it's getting worse every year.

(I have nothing against old people, I love them, they're great, I think they're actually very underappreciated by the young. But come on, know when to step back and become a mentor and not a boss. Mike Tyson should've never taken that fight and you should've never applied for that R01 grant.)

The reason we still have no idea what's going on with Alzheimer's, after decades upon decades of research, is because the NIH enabled an entire field based on lies and fake data. This has been well-documented by publications like Stat

.

For example, one of the biggest researchers in the field became a senior NIH official and oversaw a $2.6 billion/year budget until a few months ago when it was discovered that he faked data in at least 132 research papers. How is it even possible to publish so many fraudulent papers and for every single person in the entire field not notice this for decades?

The NIH world is the world in which the majority of cancer papers are fake, NIH's Cancer Institute ($7b/year budget) fails to report data on time for the cancer clinical trials it runs a staggering 70% of the time, and yet a researcher who cured herself of cancer couldn't publish her results and is being castigated by the establishment for lack of "ethics".

This is why "believe science" has turned into a stupid joke.

Despite all of this, the NIH will simply not accept any kind of oversight.

In 2006, the Congress mandated a creation of the Scientific Management Review Board at the NIH that was specifically created to provide legislative oversight into the activities of the agency and to suggest improvements to its operations. In 2015, the NIH simply abandoned the committee and has not convened it ever since.

Earlier this year, The New York Times reported on how the NIH's senior staff in the office of Anthony Fauci, including his senior advisor of 24 years, committed to a campaign of destruction of any evidence that might show up in the FOIA requests related to the agency's role in Covid-19 ("I learned from our foia lady here how to make emails disappear after i am foia'd but before the search starts, so i think we are all safe") (more details).

I don't care whether you think covid came from bats or leaked from a lab. It's the dedication to escape public oversight, combined with any lack of consequences for anyone involved that is staggering.

2 years ago Joe Biden tried to set up a new life sciences funding agency modeled after DARPA, called Advanced Research Projects Agency for Health (ARPA-H). The idea was to set it up completely independently of the existing funding structures, giving ARPA-H the full freedom to pursue the most ambitious projects it can find. Instead, the NIH swallowed ARPA-H up, denying it independence and ensuring that the NIH maintains its virtual monopoly on biomedical research.

A Professor friend of mine who received more than $10 million over the years from the NIH told me: "All of my most interesting work has been done in spite of the NIH rather than thanks to the NIH".

How did lying for money become a part of a scientist's job description? How did the public get used to no longer believing scientists? How did we grow so complacent to accept billions of dollars that are supposed to be pushing the frontier of science being set on fire year after year? How did we become so cynical to just accept all of this as a status quo?

We've had 70 years of the NIH. It was built for a post-WW2 world that hasn't existed for decades; it has done its job; and it has turned into something closer to a mafiosi organization than a science funding agency.

It's time for us to figure out better ways to fund science: we should gradually sunset the NIH over the next 10-15 years. And if you make the NIH stop awarding new funding to literal retirees, it'll free up about $5 billion every year to run all kinds of experiments.

For example, the government could solicit proposals for different funding agencies, each with a $1 billion/year budget and a 10 year mandate. Pick the 5 most interesting ones, fund them, study them, and then decide which models are working and which aren't, while continuing on with all of the existing NIH funding allocated to people of working age. Is there any reason on earth not to do it?

It's time to abolish the NIH. America deserves better. The world deserves better.

New Writing

• Intelligence killed genius

New Science

• Hiring Head of Program and Head of Content

• also fundraising -> alexey@newscience.org

• also, as usual — if you do biology or something related to biology, please email me and I would love to talk to you

Miscellaneous

• How I raised a $11M seed as a first-time, female, solo founder for a biotech moonshot by Celine Halioua

• Really enjoying Elliot Hershberg’s Century of Biology: https://centuryofbio.substack.com/p/completing-the-human-genome…. - he somehow manages to go from the very basics to the frontier in a couple thousand words every time. Amazing.

• Best of Twitter for June (1, 2, 3, 4)

• still thinking about this tweet:

• and about this tweet:

Have a great July…

Cheers,
Alexey

It’s finally the time to announce the culmination of my research into the structures of science, started in mid-2018: New Science!!!

New Science aims to build new institutions of basic science, starting with the life sciences.

Over the next several decades, New Science will create a network of new scientific institutes pursuing basic research while not being dependent on universities, the NIH, and the rest of traditional academia and, importantly, not being dominated culturally by academia.

Our goal is not to replace universities, but to develop complementary institutions and to provide the much needed “competitive pressure” on the existing ones and to prevent their further ossification. New Science will do to science what Silicon Valley did to entrepreneurship.

New Science is a research nonprofit incorporated in Massachusetts with 501c3 status pending. The board of directors consists of Alexey Guzey, Mark Lutter, and Adam Marblestone. New Science is advised by Tessa Alexanian, Tyler Cowen, Andrew Gelman, Channabasavaiah Gurumurthy, Konrad Kording, and Tony Kulesa.

In the summer of 2022, New Science will run an in-person research fellowship in Boston for young life scientists, during which will collect preliminary data for an ambitious idea of theirs. This is inspired by Cold Spring Harbor Laboratory, which started as a place where leading molecular biologists came for the summer to hang out and work on random projects together, and which eventually housed 8 Nobel Prize winners.

The plan is to gradually increase the scope of projects and the number of people funded by New Science, eventually reaching the point where there are entire labs operating outside of traditional academia and then an entire network of new scientific institutes.

In the process, we intend to both enable researchers who would’ve been working in traditional academia to work on problems they could not work on in academia and to increase the absolute number of people who work on pushing the frontier of science, by attracting those who want to pursue basic research but would not have chosen to pursue a career in traditional academia.

New Science's 2021 plan is to:

1. Prepare to run the summer in-person research fellowship for young life scientists to work on exploratory research projects they couldn’t work on otherwise in 2022

2. Continue to dig into how exactly the structures of science work and publish the results of that research

3. Get 501c3 status and raise funding for the first year of operation

If you'd like to learn more about New Science's next steps and/or are interested in:

1. Joining New Science and helping to build the new institutions of basic life sciences

2. Supporting New Science financially

3. Taking part in the summer fellowship mentioned above as a student, mentor, organizer or otherwise

4. Or getting involved in some other way

Please reach out.

And let's make science advance one young scientist at a time, not one funeral at a time.

For more background thinking, check out https://newscience.org.

Cheers,
Alexey

Writing

• In November, I published The most we can say about earnings of Substack's top writers

• If you read the paper “Are Ideas Getting Harder to Find?” (pdf) you might be interested in reading a draft of my new critical essay about it. Lmk if so.

100 of my best tweets of 2020

are here. Some examples:

Quote of the month

is from Infinite Jest:

He had tried to stop smoking marijuana maybe 70 or 80 times before. Before this woman knew him. She did not know he had tried to stop. He always lasted a week, or two weeks, or maybe two days, and then he’d think and decide to have some in his home one more last time. One last final time he’d search out someone new, someone he hadn’t already told that he had to stop smoking dope and please under no circumstances should they procure him any dope. It had to be a third party, because he’d told every dealer he knew to cut him off. And the third party had to be someone all-new, because each time he got some he knew this time had to be the last time, and so told them, asked them, as a favor, never to get him any more, ever. And he never asked a person again once he’d told them this, because he was proud, and also kind, and wouldn’t put anyone in that kind of contradictory position. Also he considered himself creepy when it came to dope, and he was afraid that others would see that he was creepy about it as well. …

He pulled his necktie down smooth while he gathered his intellect, will, self-knowledge, and conviction and determined that when this latest woman came as she surely would this would simply be his very last marijuana debauch. He’d simply smoke so much so fast that it would be so unpleasant and the memory of it so repulsive that once he’d consumed it and gotten it out of his home and his life as quickly as possible he would never want to do it again. He would make it his business to create a really bad set of debauched associations with the stuff in his memory. The dope scared him. It made him afraid. It wasn’t that he was afraid of the dope, it was that smoking it made him afraid of everything else. It had long since stopped being a release or relief or fun. This last time, he would smoke the whole 200 grams — 120 grams cleaned, destemmed — in four days, over an ounce a day, all in tight heavy economical one-hitters off a quality virgin bong, an incredible, insane amount per day, he’d make it a mission, treating it like a penance and behavior-modification regimen all at once, he’d smoke his way through thirty high-grade grams a day, starting the moment he woke up and used ice water to detach his tongue from the roof of his mouth and took an antacid — averaging out to 200 or 300 heavy bong-hits per day, an insane and deliberately unpleasant amount, and he’d make it a mission to smoke it continuously, even though if the marijuana was as good as the woman claimed he’d do five hits and then not want to take the trouble to load and one-hit any more for at least an hour. But he would force himself to do it anyway. He would smoke it all even if he didn’t want it. Even if it started to make him dizzy and ill. He would use discipline and persistence and will and make the whole experience so unpleasant, so debased and debauched and unpleasant, that his behavior would be henceforward modified, he’d never even want to do it again because the memory of the insane four days to come would be so firmly, terribly emblazoned in his memory. He’d cure himself by excess.

Miscellaneous

• 11 new discussions on forum.guzey.com in the last 2 months

• launch of newscience.org is delayed but it’s still soon

In October I,

• published Neurodiversity, mutants, and organisational design - it’s an email I encountered a while ago that I felt compelled to share. It’s a bit cryptic but I think very much worth the investment:

  Test scores are Gaussian.

  But that’s only because they squash them onto one - grading on a curve.

  The world of mediocristan.

  Human capability follows a power law and in that case what matters most as a rule is not what your weaknesses are (provided you have found a way to survive and cover those bases) but what you can do that almost nobody else can.

• published a video of my conversation with Brian Skinner (a physicist at Ohio State University, who I know because I discovered his blog back in high school). Some of the topics we discussed:

  • 01:00 - what it's like to apply for college when you don't understand the system and don't understand whether you're actually smart or not

  • 08:00 - on the "27-dimensional vector space" view of intelligence

  • 21:30 - why reading the arXiv is a form of emotional labor

  • 23:20 - how you change personally when you transition from being a postdoc to being a professor

  • 33:00 - Brian's favorite crackpot stories

  • 45:10 - on neuroatypical people in physics and the importance of social skills and having smart friends

  • 54:00 - on getting married young

• was just totally blown away by this video: I Built a REAL-LIFE Time Machine!

• a reminder that if you do anything related to biology, please reach out, I would love to talk to you, as I continue to think about https://guzey.com/how-life-sciences-actually-work/ a lot

  A few of my October tweets:

Tweet doubling as a quote of the month:

also, launching newscience.org in November :) I’m going to be raising money to fund my metascience work and newscience.org soon, but, before I officially announce this, I figured I should let my subscribers know informally - feel free to reach out if you’re interested in potentially helping to fund me; otherwise, more details in December and January..

Have a great November!

Cheers,
Alexey