unlearned helplessness | partner @lowercarbon, previously @stripe

sf
Joined December 2011
I wrote an essay about why the next frontier for AI is the world outside the data center Nature is a system more complex than anything we've ever created AI can learn how the system fits together, so we can finally understand how to shape it How? naturalgeneralintelligence.a…
84
222
69
1,814
361,386
excited to back @kevinweil @dpatil and co
💥 Introducing Warp Sciences, a new company applying AI and robotics to help the world develop new medicines faster. We’ve raised a $215M Series A led by @lightspeedvp to bring frontier AI out of the browser and into the lab, and build scientific capacity here in America. I co-founded Warp with my long-time friend @dpatil, FAIR co-founder @manohar_paluri, and drug hunter Sol Reisberg. The early team is full of incredible people like @bkspears9, who helped achieve Q>1 fusion at LLNL.
16
928
Ryan Orbuch retweeted
Today @standinsurance announced $70M at a $1B valuation, led by @Saga_Ventures. A year ago, we insured $1B of property. Today it's $10B. Disasters follow the laws of physics. Insurance should too. Here's what our team is building:
59
24
28
179
701,878
Ryan Orbuch retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/
873
2,613
1,432
19,711
4,436,293
Ryan Orbuch retweeted
Tomorrow on Cheeky Pint: @willob of Ulysses joins @sacca and me to discuss why seagrass is the ultimate carbon sink, scaling underwater hard tech (deep tech?), and selling sea drones to the US Navy.
60
29
23
740
166,012
Ryan Orbuch retweeted
Amen to everything Ian said. Arbor is one of the most ridiculous technical teams in Lowercarbon’s entire portfolio, and when the dust settles they will be one of the defining energy companies of this era.
Is Arbor Energy the key to power AND water for data centers? They just cracked the code to supercritical CO2 turbines, which have been something of a holy grail in power generation, especially in Arbor's case, because: 1. Using sCO2 as a working fluid can achieve energy efficiency rivaling large combined cycle gas plants in a much smaller footprint––e.g. behind the meter at data centers. No one had pulled it off though because it requires engineering a system capable of withstanding immense pressure, temperature, speed, and corrosion. Until now! 2. Combining this with oxycombustion of natural gas (CH4) means the outputs are only CO2 (used, recycled, and/or sequestered) and water. Arbor's power modules will PRODUCE tons of water while powering data centers... a MILLION gallons per megawatt, to be exact. So not only will their 25MW Halcyon system power data centers, it can ELIMINATE their water burden and even produce surplus water for local communities! 3. The key components can be 3D PRINTED! You've surely heard of the bottleneck in gas turbines, which is primarily due to the way in which combined cycle turbine blades must be produced (single-crystal casting). I'm extremely proud that @cantos was Arbor's very first investor and has invested in every round since then. Watch this company!
4
6
1
59
6,572
Ryan Orbuch retweeted
Wood is the rarest material in the Universe.
SpaceX will capture an asteroid with more gold than Earth's crust and land Optimus robots carrying American flags on Mars within 8 years. Starship parks Psyche in geosynchronous orbit over an American Pacific atoll. No humans within 50 miles. Optimus robots extract the metals. Delivery to Earth is free though some burns up on entry. Bezos said 15 years ago Earth gets zoned residential as heavy industry moves to space. Vast distances make emissions irrelevant. The same fleet lands on Mars in 8 years max. Optimus robots deploy solar panels, batteries, compute, and Starlinks for 4K video before humans arrive. SpaceX moves the factories off the planet.
95
272
20
6,469
162,692
I’m surprised people aren’t using the term evolution for this, developing some notion of speciation re the particular expertise of subagents, some concept of an ecological niche within which that species performs best
You can always rely on @kevin2kelly to conjure up mind-expanding visions for the future of what machines want kevinkelly.substack.com/p/sw…
1
1
4
913
I’ve seen people talking about agent ecologies which seems directionally right, but begs the question of speciation or something like it
1
4
417
maybe @deepfates and co have something here
1
315
has anyone aggregated/denoised all this radar data on bird migrations?
If you have been paying attention to radar this morning, you may have noticed it looks quite active even though there is little to no precipitation. Much of this radar activity is actually migrating birds! 🦅 🐤 #flwx (1/thread)
3
5
1,314
Ryan Orbuch retweeted
The best place to put data centers might be the middle of the ocean. After studying and corresponding with @_panthalassa, a startup building wave-powered data centers, I’m releasing a model and interactive dashboard that tests the viability and cost of moving compute to sea. If you’re curious about what harvesting massive amounts of energy from the ocean looks like and why I believe ocean compute beats data centers on land, read it.
6
23
12
62
10,496
Ryan Orbuch retweeted
this research direction has been criminally under-discussed by serious AI people. I'm not sure why. maybe the notion of "models in sync with nature" is somehow too woo. or is seen as not serious in comparison to x-risk. in any case, I think this is a mistake that needs to be remedied. thanks to @orbuch for writing it up!! hope it garners lots of interest
I wrote an essay about why the next frontier for AI is the world outside the data center Nature is a system more complex than anything we've ever created AI can learn how the system fits together, so we can finally understand how to shape it How? naturalgeneralintelligence.a…
3
1
20
2,623
disappearing wormymorph
This quoted post is unavailable.
3
933
3blue1brown comes for Arbor Energy! xitter.cf/orbuch/status/18420450…
Ok this is a 60s version even better!!
1
985
Ryan Orbuch retweeted
Today I joined the investing team at @lowercarbon I started my career in big law and banking. I didn’t expect that path to lead to factories and rocket launches, but working alongside hard-tech founders has reshaped my thesis about life and the future. 1/n
73
8
9
317
36,677
this is like fred again we’ve lost dancing timepinning early covid but for today, it’s so good
I gave Claude another 18 hours.. and I think this one is the best one yet Macrohard: Windows XP I'm blown away
1
14
3,275
all-timer
[slams hand down on table] rationality is an attempt to reconstruct judaism but using exclusively the mind instead of the heart, and effective altruism is the same thing but for christianity. motherFUCKER
2
28
6,809
Ryan Orbuch retweeted
In a break from the current VC world drama, I pose the question everyone actually cares about. Can we use AI to talk to whales? @ProjectCETI’s David Gruber is building the infrastructure to enable the first human-whale conversation and initiate an interspecies dialogue. We discuss: - David’s journey to sperm whales - How AI can translate a language no human has ever spoken - Sperm whale culture & the whale economy - The seismic societal implications of human-animal communication - The future is going to be much stranger than we can imagine. Enjoy! TIMESTAMPS 0:00 Intro 1:48 David's Journey 12:30 An Interspecies Dialogue 17:14 Sperm Whale Language & Culture 23:48 The Breakthroughs Yet to Come 27:49 The Whales Are Yappers 31:03 Sperm Whales Have Accents 39:47 Verifying a Whale Translation 42:13 What Success Looks Like 47:32 Saving the Whales
15
10
8
102
10,169
This is excellent 🫡 Its pretty fucking cool that we get to live at the time when there is a "White House Accord on Super Intelligence." Lets work together and do this right.
7
1,297
We need equivalent of Kardashev scale for planetary system stabilization. Type I civilization is able to keep planetary systems within livable bounds. Type II civilization has fine grained control over all home planet systems and can achieve livable conditions on a new planet. Type III civilization can establish instant fine grained control on any arbitrary new planet or other gigascale structure.
40
43
2
499
1,134,230