Discovery Loop

They make many bold promises, but their core goal is neatly encapsulated on the website:

"Imagine a future where a handful of people can conduct scientific research and engineering tasks much more rapidly, and with higher quality, than massive teams of scientists and engineers do today."

This is the same goal as every other AI company out there. Automate away the human employees and let a small number of "people" (note that they do not say scientists or engineers for this part) take the credit and financial rewards for every good thing this human-free system produces.

The solution to most of these problems lies in policy, not in new tech advancements.

Maybe if our biggest companies did something other than suck up to science denying wackos, some progress could be made in these areas.

> Provide Access to Clean Water

??? We don't need any AI for this.

Start with separated sewage/wastewater and stormwater drains. Then accredited and highly scrutinised wastewater treatment and discharge into water bodies (or see below for a high-tech solution). As for clean water to the home, direct those stormwater drains to new reservoirs which sustain freshwater aquatic life. Protect aquifers from over-drainage, and build pipelines from water-abundant regions to water-scarce regions.

To reclaim waste water or treat unknown water sources back to potable/semiconductor standards we have ultrafiltration, reverse osmosis, UV treatment, pH adjustment, fluoridation, desalination, softening (which is generally obviated by RO...). This is basically Singapore's NEWater.

Good sanitation is a financial and political problem. The engineering has been solved for decades now.

> Make Solar Energy Economical

Isn’t it already?

Sandbox 2.0

But also, solar power is already economical.

  3. Develop Carbon Sequestration Methods

Would be great if they'd add:

Reverse human aging.

(Maybe a sub-topic under "Engineer Better Medicines".)

Room Temperature Ambient Pressure Super Conductors

Very honorable effort, but a lot of these seem to touch heavily regulated industries impeded by unwise or outdated policy no less than by the lack of clever engineering - medicine, education, urbanism, energy. I wonder if they've given some thought to the key blocking factor as well.

Such a weird list. How is preventing nuclear terror an engineering problem?

Why is "12. Enhance Virtual Reality" in there? T_T

Many of those require a paradigm shift in our understanding of Universe, so I am not sure they are achievable with any convex combination of existing knowledge. We'd need some sharp mind connecting the dots but with heavy use of AI and incentives to use it, we might just never give a chance for such mind to arise.

> 1. Make Solar Energy Economical

This one is already solved, right? The price of panels and batteries is on trend to displace all other forms of power generation within our lifetime

> Make Solar Energy Economical

That is already solved.

> Develop Carbon Sequestration Methods

That is not necessary, because 1 is solved.

> Reverse engineer the brain

What for? There was already the european human brain project, which didn't do anything useful.

> Prevent nuclear terror

Easy one: Every country stops developing nuclear weapons and destroys existing ones.

It seems this list itself has many flaws. Maybe we need a bigger computer which figures out the questions we really need to ask.

> 9. Reverse Engineer the Brain

For what purpose? To replace humans? To make social media more addictive? To master brain manipulation?

This list smells of so much tech bro “I can do better than the people that have been working in the field for 20 years” egotistical attitude that permeates silicon valley. Why do software engineers think they are smarter than everyone else? Is it because they earn more than most people? But then bankers and finance bros should think they are gods?

Also, solar energy is already economical!? Do they mean more economical?

In what sense is solar energy not already economical?

15. Cut AI energy use by 1000x while increasing processing speed 1000x

Obvious near term trillions dollar market to disrupt.

I would say 5, 6, 10 can be even done today if we had right politicians that can make policies for the people

I think this should and can be done openly: https://opendl.ai

1. Make Solar Energy Economical — https://github.com/orgs/HardisonCo/projects/194

2. Provide Energy from Fusion — https://github.com/orgs/HardisonCo/projects/206

3. Develop Carbon Sequestration Methods — https://github.com/orgs/HardisonCo/projects/196

4. Manage the Nitrogen Cycle — https://github.com/orgs/HardisonCo/projects/197

5. Provide Access to Clean Water — https://github.com/orgs/HardisonCo/projects/195

6. Restore and Improve Urban Infrastructure — https://github.com/orgs/HardisonCo/projects/193

7. Advance Health Informatics — https://github.com/orgs/HardisonCo/projects/203

8. Engineer Better Medicines — https://github.com/orgs/HardisonCo/projects/200

9. Reverse Engineer the Brain — https://github.com/orgs/HardisonCo/projects/205 10. Prevent Nuclear Terror — https://github.com/orgs/HardisonCo/projects/204

11. Secure Cyberspace — https://github.com/orgs/HardisonCo/projects/201

12. Enhance Virtual Reality — https://github.com/orgs/HardisonCo/projects/202

13. Advance Personalized Learning — https://github.com/orgs/HardisonCo/projects/199

14. Engineer the Tools of Scientific Discovery — https://github.com/orgs/HardisonCo/projects/198

Specs and the per-challenge process lists: https://github.com/HardisonCo/opendl

It should also be funded by the Gov. IMO and 100% for oublic benifit e.g.: nsf.dev

Which engineering discipline touches most of these?

Please add fixing neuro issues like autism add etc on the list. It creates a huge burden on families.

Sounds a bit like everything and anything to be honest. Focus?

15. Unscramble an Egg

16. Make Everyone Nice

17. Finally Impress a Girl

1. Develop AGI

2-14. ???

Acquisition back by Google in 3 years, with nothing to show for it. VCs will make a ton.

1. Make Solar Energy Economical - Disallow fossil fuels

2. Provide Energy from Fusion - See 1

3. Develop Carbon Sequestration Methods See 1

4. Manage the Nitrogen Cycle - See 1

5. Provide Access to Clean Water - See 1

6. Restore and Improve Urban Infrastructure - See 1

7. Advance Health Informatics - See 1

8. Engineer Better Medicines - See 1

9. Reverse Engineer the Brain - See 1

10. Prevent Nuclear Terror - See 1

11. Secure Cyberspace - See 1

12. Enhance Virtual Reality - See 1

13. Advance Personalized Learning - See 1

14. Engineer the Tools of Scientific Discovery - See 1

FF is the real threat in time, money, health. Can't sweep aside that it will destroy most life on Earth and we'll never get to the other things if we are at the mercy of FF

This reads as hype of the kind: "well, we can't reliably do these rather mundane things with AI, but we're going to run it extra hard and extra long in some novel way, and it will do amazing things".

The issues are with verification and with detecting drift from the goal. These are related, if not roughly the same issue. And, if they can solve this, then they will have essentially fixed AI. Maybe even AGI.

But, if this were the goal, then it seems more reasonable to solve the relatively more mundane verifiable challenges (e.g. generating solid, reliable code). Then, working up from there.

And, that's exactly what gives this the hype smell. No use for solving problems that don't get the oohs and aahs. Just straight to NAE Grand Challenge problems.

dillusional millionaires

- Still doesnt solve the core problems

- Eliminate racism

- Eliminate poverty

- Eradicate crime

- Eradicate corruption

- Reverse climate change 100%

- wake me up when you got an AI project capable of doing this one

What solving these challenges benefit us? Some of them make sense, but enhance virtual reality? Reverse engineer the brain? Yeah, good luck with that.

4. Manage the Nitrogen Cycle

Easy solution - eat less products that pass an animal first - reduces nitrogen pollution by 10x intantly, low tech.

I'd re-formulate: 4. Make people more flexible to changing their mindsets & habits - this is the ultimate problem.

What is the most impressive thing Jeff has ever done?

I only see "co-created" "co-founded" "managed a team" - what did he actually do?

The risk is they are a magnet for many more talented ML scientists to leave google.

It depends on your lens, some may think it's a way to show them the doors after failed corp politics.

[dead]

That's a very silly comparison, there are many startups working on RSI, karpathy is just a basic version to try the concept (similar to his gpt work)

That is also what I understand from that page. Autoresearch is the closest thing we (mainstream audience) know of but I am sure there is already an active research literature around it.

this is like saying taylor swift must have been influenced by justin timberlake because they both dance on stage sometimes.

In my area (pharma) what it looks like is this: A human defines a high-level research objective. "Identify a protein target that causes disease in humans, and find a molecule that binds to, and disables, that protein, eliminating the disease".

That objective then gets loaded into an ML model that spits out an experimental protocol. A protocol can be as simple as: "make 1 million test tubes, each with the protein, and in each, a custom molecules, and look for test tubes that show some reaction of interest". It can be a lot more complicated (for some reason, biologists who run these systems always try to do the most challenging experiments first, while I tend to spend all my time demonstrating the system can pass basic controls first). The protocol is then loaded into a robotic work cell which has access to protein-making machines and drug making machines, and then it handles all the experimental details (which previously would have been done by a technician). It scales up far larger than individual technician, is much more reliable, and faster (in theory- all of these are aspirational goals right now). T he results of those experiments are used to fine tune the experimental protocol and run another round. You run this in a loop and the result is better drugs faster (again- in theory.)

This is already an active area of research with more resources going to into it every day. The fact that Jeff and Sanjay have chosen to bet on this approach should be no surprise. In many ways, this is exactly what I intended when I wrote the documents inside Google (15 years ago) that motivated Jeff and Sanjay to work on scientific computing problems, and my current company is already trying to figure out how to work with Discovery Loop.

would love to see how AI can automate the construction of the next high energy particle collider

You're halfway there, but the only impediment isn't on the side of the researchers. Many of these topics they're trying to solve involve human subject research. Even with tireless embodied researchers who work around the clock and don't require breaks, you can't make the thing you're studying happen faster. The biggest reason we use poor proxy measures for things like longevity and mortality research is the simple impracticality of finding two groups of randomly selected people, ensuring you can control their entire lives for 60 years, the only difference between them is one variable, and see who lives longer. Putting aside the ethics, even if you could find willing subjects and actually control their entire lives to that extent, it would still take 60 years to gather the data you need. It doesn't make any difference whether robots or humans are running the program.

One of my favorite books from the past few decades is The Extravagant Universe, written by one of the astronomers who helped discover dark energy and develop the current most-accepted model of cosmology. I love this book because of the emphasis on physical process in astronomy. Part of the reason it took decades to study this problem is they need to collect data from supernovae. Those only happen so often in places we're looking. You can't automate alignment of the heavens. It happens when it happens.

> transcendence

> immanence

somebody has been studying Christian theology!

You can use simulators. However the problem is that if you're for example running material science experiments, those simulations will consume a lot of compute and take weeks, so spamming different approaches in the way an agent tends to work might not work quite as well.

A public benefit corp shouldn’t be a start up. The primary goal of a start up is to grow as quickly as possible which is rarely benefits the public.

Very silly to call every non start up a lifestyle business. It’s just a business. Start up are the weird thing that almost always an obscene waste of time and money, but sometime creates google.

Yeah, the reason I don't see it how it would be successfull is that most public labs are usually struggling with money, so it would definitely cost them less to build their own automated pipeline.

It can be both. Bell Labs performed a lot of speculative research while still producing economically valuable technology.

> To be honest, this feels more like a lifestyle business (aka hobby) than a startup

They're also incredibly productive and can build/deliver really good stuff, so who knows :)

I don't see how not? A theoretical physicist can do all the thinking they want but if they can't test an idea against nature it's not super useful.

> Our mission is straightforward: we are building AI solutions that can automatically solve important problems in machine learning, science, and engineering.

Genuinely curious which part you found complex.

Right. What about the scientific hardware (instruments, sensors, robotics)? Partnerships with existing research institutions? Dealing with restricted data?

Modernizing science is a lot more complicated than just optimizing the inner experimental loop, but their hiring page implies it's a pure ML lab focused mainly on model development.

Which part of it is highly technical or jargon loaded?

It’s true in the limit, but I think we are nowhere near that limit. A friend recently started a bio startup and automated parts of mouse experiments enabling higher throughput on in vivo experimentation. This is not commonplace, and there is a lot of room for further automation here.

As anyone who works with agents daily can attest, 1) you can use agents to help with hypothesis refinement, bridging into areas adjacent to your expertise, etc. 2) once you have a rigorous /goal definition you can parallelize and let the agent crank.

It seems pretty obvious to me that with the right actuators and sensors you can apply this to real physical research loops too. (To be clear, this is not easy; a lot of bench work is Métis and needs experts in the loop at every stage.)

To your point, you can’t make plants grow faster but you can increase research throughput by enabling a researcher to have 10x or 100x as many experiments going at once.

This is a valid take but at this moment, AI is not trying to solve this fundamental dynamic. But it can still accelerate the process by aggressive exploration of the solution space which cannot be done even with an army of human researchers. Many ideas can be relatively quickly verified (and discarded if needed) by proper simulation even before real world experimentation, but we don't have enough capacity to process all potential ideas. If you can build a good model for simulation and establish a robust methodologies, we can use some ideas which never had a chance before.

> It might be a new scientific revolution to have computer-driven discovery.

And ... it might not.

> Note that Jeff and crew have cleverly structured their company to avoid problematic uses of AI

Do you have more sources/info on this?

Like how OpenAI is (was?) structured as a nonprofit?

There's this somewhere on that page:

> securing cyberspace,

which has clear military implications, at least in today's age.

I'm a scientist. On the one hand I take some comfort in thinking that I will always have an advantage in the lab. On the other hand I'm not taking anything for granted. And my advantage in the lab has to translate into an employer being smart enough to keep me around until if and when the AI takes over, which kind of translates into their investors wanting to keep me around.

We know what happened to manufacturing when investors were no longer interested in it.

I’m not sure that I agree entirely with your framing here, yes, you do at some point need to correct your assumptions with external evidence. But clearly there have been many individuals throughout history who have had incredibly out sized impact in their respective fields as a consequence of the quality of their reasoning.

Would you side with the adjacent statement "expertise is not the bottleneck"? Didn't we recently discuss "LLMs reward expertise"?[1]

I suspect Discovery Loop will have to hire experts in each area they are targeting, to supervise and prompt their system effectively, much like the Terence Tao conversation with ChatGPT the OP cited[2].

[1] https://news.ycombinator.com/item?id=49161518 [2] https://www.seangoedecke.com/llms-reward-expertise/

I don't see this group needing funding, in fact I doubt you could get a meeting to try to invest unless you're a personal friend.

«Jeff Dean's PIN is the last 4 digits of pi.»

I had not read this before, but told many students the same about my PIN code and I a quiz about the last digits. Love it.

Those are all fake, part of an internal Google narrative that overstates individual contribution, and obscures the work of large engineering teams.

Here are some Jeff Dean well sourced facts:

- Already part of engineering of Google indexing systems that lacked basic checksums and ran on non-ECC hardware, allowing silent data corruption.

- One of the authors of LevelDB a database with so many documented crash-consistency, recovery, and data-loss weaknesses for years. Just check their Github project. LevelDB current tracker contains unresolved crash consistency, recovery and corruption reports going back almost 12 years on GitHub

- In AI engineering technical lead, let TensorFlow lose researcher mind share to PyTorch, and caused Google fragmented landscape across TensorFlow and JAX.

- Had the people at Google who invented the Transformer architecture, but failed, to turn that lead into the first dominant public LLM.

- As AI engineering and VP management let Google Brain and DeepMind remain duplicated and internally competitive for too long.

- Let Noam Shazeer leave and then spent heavily to bring him back with nothing to show for.

- Part of Technical VP leadership who had Bard rushed to launch with factual errors in Google own promotional material.

- The first Gemini demonstration overstated how real-time and interactive the system actually was, being basically a fake.

- Part of the VP and AI technical leadership who had Google AI Overviews launched with weak source quality controls and repeated satire and low-quality web content as factual advice.

- Part of teams that launched AlphaChip performance claims that were difficult for outside researchers to reproduce and remain technically disputed.

- Jeff Dean public explanation of Gebru departure was contested and damaged confidence in Google scientific governance.

- Jeff Dean was part of the team at Google that removed or marginalized prominent internal AI ethics critics shortly before many of their warnings became product problems.

- Jeff Dean was one of the managers behind Project Dragonfly supporting censorship.

- Jeff Dean is part of the VP technical leadership approving Project Nimbus supporting an ongoing genocide.

If you check out some sub-tweets from people in the org, it wasn't really all butterflies internally for a while. Sorry, really don't want to name people and give examples.

> Automating AI research is terrifying.

what why?

> Between us, we have pioneered massive scale computing and led the creation of critical infrastructure, products, and foundational AI advances that the world relies on, including multiple generations of Google Search, Google Ads, Google News, Google Translate, Google File System, MapReduce, BigTable, Spanner, TensorFlow, Pathways, TPUs, AlphaChip, AlphaStar, AlphaCode, AlphaFold, Gemini, model distillation, mixture-of-experts model architectures, word2vec, sequence-to-sequence models, chain of thought reasoning, neural architecture search, and multiple generations of Large Language Models (LLMs) among others.

Not a bad combined CV.

[dead]

Gemini has done absolutely nothing for me. I can't even shut off the navigation feature on my phone using only hands free, when I get close to my destination. I have to take my eyes off the road, look down, and tap to exit.

Google's advanced AI cannot even exit a mobile app.

> only works for a very narrow definition of what science is

And so does academia. It's just that instead of AI and robotics, PhD students are thrown onto problems that are in large parts slightly tweaked reconfigurations of similar experiments.

Especially in chemistry, biochemistry, material sciences there is a large space of discoveries that are barely "novel" in an intellectually stimulating way, but still highly valuable that can be explored orders of magnitudes faster than is currently the case.

Lets keep your comment out of the VC pitch deck shall we?

> would an ML optimization loop have discovered transformers?

I think that’s exactly the kind of problem this group is looking to solve. You make a compelling intuitive argument, but that’s not the same thing as a proof

What percentage of people work at a startup though? Not just new/small business, which could include restaurants, local services, etc., but tech/science startups that would meaningfully benefit from AI.

I'd bet you could 10x the number and still be in low single digit percentages of the US workforce. And it seems pretty likely that AI-enabled startups will also employ less people per-startup.

If AI causes a white-collar jobs apocalypse, I don't think startups are picking up the slack, although it'll plausibly cushion the blow somewhat for top-performing tech workers.

They kind of touch on this:

> Imagine a future where a handful of people can conduct scientific research and engineering tasks much more rapidly, and with higher quality, than massive teams of scientists and engineers do today. By automating the loops of discovery, the world will be able to make much more rapid advances across countless fields of science.

The problem is that for this to actually become true, compute needs to become commodity again, otherwise this capability will select for people and environments with oversized pockets.

Why are people so sour about this?? I can read the site easily, its clear, performs well on mobile, what else do you want? Why is so offensive to people that models trained on tailwind or whatever?

If their goal is to automate scientific discovery, why would they not automate building their website?

(Though, I do wish people would use just a few extra prompts to break out of the 'vibe-coded' look.)

let me rephrase that:

"The site itself demonstrates the team is spending their money in the places that matter, and using quick solutions for the stuff they need but isn't mission critical"

At least it isn't dark purple.

Absolutely reeks of Claude. This is the new aesthetic.

My first thought when I see the website too

We'll adapt. So far, it seems like every technological increase has demanded more cognitive ability, not less. Sure, we may not need 100 people who are experts at shoveling, but the people developing, maintaining, and even operating the excavators all have to be "smarter". All the thinking, learning, and training needed to stay alive in the cold might have been mostly lost when fire was controlled, but it comes with its own risks and learning curve.

I know AI is "smart", so it might hinder us there, but I doubt it can be as damaging as doomscrolling has been on human brains.

I assume that, like most really trendy companies, they hire mostly by internal referral.

You can create a research twin that can predict but as you say, it only works once you have a lot of data. Both the data relevant to the experiment it came from (which is rarely published) and the complete failure data (which is almost never published in any form).

I truly believe if we took a measely $50b out of the LLM world we could create trillion dollar economies from basic research within 10 years. I personally know folks who have intuitive understanding of things that can't get funding to be studied. If we could keep the money away from university upper management, it'd cost $10b max.

Oh and while we're at it, $20b a year would house every homeless person in the US - there's a hell of a lot of extremely high intelligence and low social cohesion folks who can't handle the extractive punitive system we have. Our ability to deliver opportunity to create lucky situations for ourselves is getting worse and worse

Yes! But if science is bottlenecked by funding, making it cheaper might help?

LLMs can't materialize funds or political will so let's stick to running GPUs hot and publishing papers. The citations will be amazing. /s

Thats why you simulate e coli at 30x in a sim environment that fable slopped together. Duh

Google is backing it.

Google down $160Bn so far since the leaving announcements. Those are some valuable people!

I agree, for better or for worse.

If I had to bet my money, it would be on "for worse".

No it won't, not in any real world touch grass, walk around downtown kinda way. The internet will have changed but so what

It will not be owned by top 1%?

As always, a very good presentation.

I don't see the salary (on mobile). Isn't there a law stating it must be added?

I don't know why you're being downvoted.

This is basically something scientists have been alarming about for the past year: We're moving into a future where science may be tiered into the haves (those with access to premium compute) and the have nots (hoi polloi with restricted access), which in turn could seriously influence what kind of science we'll get.

Worst case, we'll get science that is completely dependent on business and politics.

EDIT: I should note, this comment was aimed at a more general case.

No its not. Everyone dies. Healthcare is invoked in everyone's life. Generally the longer you live the more healthcare you'll need.

Numerical mathematics

Does this mean anything besides for corporate virtue signaling?

Ah following in the foot steps of OpenAI. Five years later: More like Discovery Knot amirite

This is their version of retirement. Did you think Jeff Dean was going to stop doing tech and play golf?

I suspect they want a bit more control over what they're working on. These days the bean counters are running Google.

I mean what are they doing right now at Google? Optimizing data centres? Pretty lame compared to this. Even if they completely fail, i'm sure there'll be good lessons.

Why? Science is wildly unprofitable on the scale of an individual private firm.

The company is developing an application, or a class of applications. Not a new model.

I think Google's branding was starting to be too poor in AI to get top talent, they needed the refresh

Model routers - send all of your data through a third party who totally swears not to peek at it.

If youre doing anything high value (advanced research, classified work, high value industrial research, health data) then sending your data through a third party like that is insane.

He was born in the US not India, good food/nutrition must have played the part :)

> It hasn’t made me a millionaire, but I haven’t lost money either - so that’s some sort of win, right?

I'm curious if that is before or after token costs?

Don't leave us guessing... Why is yours better/different?

Google already invested into them.

It is targeted at a dozen or so people at OpenAI and Anthropic, not you or me.

I have no doubt that internal Google friction is one of the reasons they are moving. But the Gebru incident was almost seven years ago now. It is very unlikely to be a proximate cause.

I doubt numbering vs names on TPU releases even crosses Jeff's radar. It's not the kind of thing he cares about.

I'm almost certain the goal of this startup is to make physical automated research labs guided by RL

How is that different than video input?

Do you actually think that this group of founders will have trouble finding domain experts wanting to work with them?

in AI/ML, no. They are just going to automate AI/ML research to start with. Totally doable.

For some of the other things, undoubtably yes.

yes! and giving current scientific research on reading efficiency pointing towards fixed pages being much better than scroll, the whole web is screwed but these fancy sites... a big merda

You know when the page has all-caps "01 — THE APPROACH" that it is slopified. I guess I shouldn't be astounded, but I am, that world-class talents with world-class backing are just taking default LLM output and saying, "okay looks fine".

Yeah what these guys are mainly known for is vaporware

> we have pioneered massive scale computing and led the creation of critical infrastructure, products, and foundational AI advances that the world relies on, including multiple generations of Google Search, Google Ads, Google News, Google Translate, Google File System, MapReduce, BigTable, Spanner, TensorFlow, Pathways, TPUs, AlphaChip, AlphaStar, AlphaCode, AlphaFold, Gemini, model distillation, mixture-of-experts model architectures, word2vec, sequence-to-sequence models, chain of thought reasoning, neural architecture search, and multiple generations of Large Language Models (LLMs) among others.

Jeff Dean invented MapReduce and Bigtable.

Say what you want about their later years, but Dean and Ghemawat deserve their title of engineer.

Before Jeff Dean was a "manager", he got a > Ph.D. in computer science from the department that would become the Paul G. Allen School of Computer Science and Engineering at the University of Washington in 1996, working under Craig Chambers on compilers[8] and whole-program optimization techniques for object-oriented programming languages.[9]

https://en.wikipedia.org/wiki/Jeff_Dean?#Early_life_and_educ...

So he's got no shortage of talent as an IC.

Sounds like you get your news from 2024 when people thought things were plateauing after GPT4?

we definitely haven't hit plateau yet. I think a lot of people latch on to anti-LLM narratives without really thinking things through.

LLM coding isn't even close to plateauing. Right now, the major players are in a consolidation step, focusing more on economic efficiency but still not at the point where we're ready to start burning models to hardware and freezing the line.

They are straddling the line between pushing it forward, and justifying the business case. It's hard to do both at the same time.