Building @getduodata. Founded @teralytics. CS @ ETH Zurich.
Berlin, Germany
Joined November 2007
- Tweets1.6K
- Following1.6K
- Followers569
- Likes496
Georg Polzer retweeted
We were taught that Earth orbits the Sun in a closed circle. In reality, Earth has never returned to the same point in space once in 4.5 billion years.
Right now, the Sun is carrying Earth and every planet around the center of the Milky Way at 230 km/s (828,000 km/h), more than 18,000 kilometers every 80 seconds.
Because the Solar System is tilted 60.2° to the galactic plane and bobs up and down through the galactic disc roughly every 30 million years, the planets trace open 3D corkscrew helices across deep space rather than flat loops.
The last time the Sun stood where it stands today, one Galactic Year (230 million years) ago, the very first dinosaurs were only beginning to appear on Earth. Since forming 4.6 billion years ago, our Sun has completed only 20 laps.
Georg Polzer 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/
Georg Polzer retweeted
"Europa przegrywa z Chinami, fabryki upadają"?
Oto twarde dane z gospodarki strefy euro:
1. Prawie 2/3 importu z Chin to zaledwie 5 branż (chemia, auta, proste maszyny).
2. W tym samym czasie unijny przemysł farmaceutyczny, optyczny i elektronika specjalistyczna notują rekordowe wzrosty.
Efekt? Zatrudnienie i realna wartość dodana w przemyśle UE w ciągu dekady... wzrosły.
3. Tani wsad z Chin obniża koszty europejskim fabrykom (badanie EBC: +0,6 p.p. do rocznego wzrostu produkcji).
4. To Chiny są w ponad 80% uzależnione od UE w 41 strategicznych łańcuchach dostaw (od litografii ASML i optyki Zeissa po medycynę).
Chiny nie niszczą europejskiego przemysłu ale wymuszają 'twórczą destrukcję' i ucieczkę w segmenty najwyższych marż.
Prawdziwym problemem Europy nie jest Pekin, tylko leżący w szufladzie raport Draghiego (wdrożono ledwie 15% zaleceń).
Georg Polzer retweeted
What’s the evidence on rent control?
I charted the findings from 112 studies:
Great for rent-controlled tenants ✅
But:
Pushes up rents for other people ❌
Pushes down new housing construction ❌
Pushes down building quality ❌
Georg Polzer retweeted
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.
Georg Polzer retweeted
Mustafa Suleyman's whole piece criticizing Anthropic, and the way it's imbuing Claude with some sense of self-hood is really worth reading IMO mustafa-suleyman.ai/a-warnin…
Georg Polzer retweeted
ASML: "We are not selling anything in Europe at all. That is because Europe is not investing, and no chip factories are being built here". nltimes.nl/2026/09/22/asml-c…
Georg Polzer retweeted
Ali Ghodsi: “There’s no agentic work that’s automated the whole enterprise. That has just not happened.”
“I think that the real reason, if you actually look at it, is the models are smart enough, but they just don’t have the context that exists inside of any organization.”
“We don’t need a smarter model that can actually solve Navier-Stokes or conjectures or do better on Humanity’s Last Exam. We need it to just go from 60 to 70%.”
Co-sign. We’re very early on getting the most out of context.
Brilliant conversation @alighodsi @martin_casado
Databricks' @alighodsi on AI risk and adoption:
Ali isn't losing sleep over the existential risk debate.
He says a number of conditions would all have to be true simultaneously to enable an actual runaway takeoff scenario, and currently several opposite conditions exist.
Each frontier training run requires significantly more resources. Power, GPUs, engineers - and some attempts fail, burning up huge piles of money with them. Until that reverses, he doesn't see the self-improving loop happening.
Cyber is what he's watching most closely and where he anticipates real impact. Most orgs are not equipped for the coming change in agentic capabilities. The time between a vulnerability being published and being weaponized has collapsed from years to hours.
On adoption, he believes most companies don't need a smarter model. The models are already smart enough. The gap is context they don't absorb - the things an employee who's worked at a company for five years learned by osmosis.
If the frontier stopped advancing today, he thinks it wouldn't meaningfully change the value most are extracting from AI anyway.
In conversation with a16z's Martin Casado and Sarah Wang:
00:00 Intro
00:48 Why Ali places the AI risk near zero
05:05 The word "pacing" was a mistake
10:50 What 10k agents and $100m can do
12:20 What would change his mind on AI risk
14:20 More GPUs, more ways to fail
18:05 US export controls on PlayStations
20:10 The damage everyone expected by now
24:30 Public vulnerabilities weaponized in hours
30:15 Why labs can't grade each other
37:15 Why most of RSI isn't actually RSI
40:05 Why nobody really needs a smarter model
41:50 The AI use cases nobody argues about
47:30 Google Search solved this 25 years ago
50:55 Nobody has privileged knowledge now
55:10 Same model, new harness, 2x cost
58:15 Open source: 5% of spend, 60% of tokens
1:05:30 90% of new databases are created by agents
YouTube: youtu.be/GzEtpAKYRvE
@databricks @martin_casado @sarahdingwang
Georg Polzer retweeted
PETER THIEL CALLED IT.
The Antichrist wouldn’t build the doomsday machine.
This clip aged terrifyingly well.
Peter Thiel predicted in June, 26, 2025 exactly what is happening right now with Anthropic, Coxon and the AI-safety landscape.
“The way the Antichrist would take over the world is you talk about Armageddon non-stop, you talk about existential risk non-stop.”
The 'Antichrist' wouldn’t seize power by creating an AI apocalypse, but by CONVINCING everyone that the apocalypse was coming and then offering safety in exchange for CONTROL.
A must watch.
Georg Polzer retweeted
REMINDER: They need a credible off ramp from the hyper scaling narrative they’ve been selling for months. Safety concerns are an obvious way to justify the incoming pivot. They also justify taking the whole thing into a closed “Manhattan project” type setting at the government’s expense. This justifies the incoming lockdown.
As I predicted in August the scale narrative had run its course due to kimi. Hyper scaling wasn’t about genuine demand expectations. It was about winning the ai race by outspending and bankrupting your competitors. Kimi made that pointless.
This is no doubt the asymmetric information Bessent has been alluding to for weeks. There’s going to be a helluva lot of liquidity freed by this move. It’s all going to go flooding into USTs. Happy D day bond guys.
👇
Great observation from Tracy below. But I’ve annotated the chart as I think a lot of people still don’t understand that it was Deepseek that triggered this acute phase of hyper scaling.
By going open weight Deepseek commoditised AI ensuring the “there can be only one” race had no choice but to turn to capex spend instead.
It’s been a simple “who has the ways and means to moat up most expansively?” ever since. If you’re confused by the economics, it’s because the economics aren’t supposed to make sense. The irrationality of the infrastructure spend is the logic. We’re in the world of Highlander.
That means you win by forcing everyone else to go bankrupt. Potentially even yourself (safe in the knowledge that if you do Uncle Sam still inherits the legacy infrastructure.)
In that sense it’s no different from the USSR diverting the nation’s entire consumer budget into vast steel plants that no one else could ever afford to match on cost.
But if loss leading your way to AI domination through control of compute was the plan, last month’s Kimi moonshot disrupted everything. It’s like beating Russian steel with the invention of a Star Trek replicator.
That’s why I’d be surprised if the pace of debt issuance continues on at the projected pace. If anything, Leopold’s blow up implies a pivot towards greater capital efficiency is incoming.
Alternatively this is when “the project” begins in earnest.
Either way, I’d bet on more secrecy and even more centralisation.
The only question now is whether the USG takes control or not. Judging by Anthropic’s previous spat over Pentagon access, this is the point where the statecraft is gonna get complicated.
👇
Georg Polzer retweeted
The Euro lives on German altruism, but - as Germany falls on hard times - that altruism is fading and means the future of the Euro will be increasingly up for debate in Germany. I talked about all this on a live stream a few weeks ago. You can watch here:
robinjbrooks.substack.com/p/…
Georg Polzer retweeted
Housing construction tracker for Berlin, pretty cool website. Shows blocked and delayed projects.
wohnungsbautracker.de/
Georg Polzer retweeted
Wenn man mit 70k Spitzenverdiener ist wieso kann man sich dann trotzdem kein Haus leisten
Wenn man mit 70k Spitzenverdiener ist warum wartet man dann trotzdem 8 Monate aufn Facharzt
Wenn man mit 70k Spitzenverdiener ist warum kann man sich trotzdem nur n kleines Auto leisten
Die Bundesregierung hat die Reform des Einkommensteuergesetzes beschlossen. Die Spitzensteuer soll erst ab einem zu versteuernden Einkommen von 70.600 Euro fällig werden. trib.al/audlLcn
Georg Polzer retweeted
The "water waste" from this bag of 900 almonds is the equivalent of about 34.2 million queries on ChatGPT (according to @sama).
You could personally do 1,000 AI queries per day on the current models for the next 80 years and you won't waste as much water as that bag of almonds.
