Clusters or Chaos
Mapping the geometry of LLM failure modes (NTNU, in progress)
Hi, I'm Jørgen
Former national team athlete, now building at the edge of AI and quantitative systems.
I chose breadth on purpose. Every summer a different industry, every year a different problem. Basketball, medicine, mathematics, sovereign wealth funds, startups. Not because I couldn't pick, but because range is how you develop taste, and taste is what matters when everyone has access to the same tools.
The most important skill of the next decade is the ability to think differently from the machines you work with. AI makes output converge. The more you rely on it, the more your thinking starts to mimic it. The edge belongs to people who bring variance: unexpected connections, contrarian instincts, ideas a model would never reach. In a world where anyone can build with AI, small teams of sharp, different people are the ultimate multiplier. Together you arrive at ideas no individual and no model could reach alone.
Projects
A curated selection of work across quantitative finance, AI research, and product.
Watching
I Solved Klotski
2swap
How They Got Rich!
Patrick Boyle
How Long is Now? Introduction to Metaphysics
exurb1a
You'd Be Surprised How Bad of a Person You Are
Pursuit of Wonder
Why The Rich Should Decide
Art of the Problem
Amazon Created The Socialist Dream
Art of the Problem
The Philosopher Who Started a Cult
Sisyphus 55
Pleistocene Park: Reviving the Mammoth Steppe
Atlas Pro
Can Humans Make AI Any Better?
Welch Labs
imec: The Most Important Company You've Never Heard Of
High Yield
Machine-Assisted Proofs
Simons Foundation
The Death of Socrates: How To Read A Painting
Nerdwriter1
Death Is a Problem We Refuse to Solve
CGP Grey
The Dark Matter of AI: Mechanistic Interpretability
Welch Labs
Jeff Dean Says AI's Biggest Opportunity Is Still Untouched
Laude Institute
The Kelly Criterion
Mutual Information
How To Think About Thinking Models
Neel Nanda
The Napoleonic Wars (Part 1)
OverSimplified
A Quant's Perspective on the 2008 Financial Crisis
Susquehanna
A* Search: How Map Apps Find Shortest Routes
Reducible
The World's Most Important Machine
Veritasium
The Worst Horror On Earth You Aren't Being Told About
RealLifeLore
The Hardest Problem on the Hardest Test
3Blue1Brown
I Implemented Gaussian Splatting in Unity
IndividualKex
Advice for Time Traveling to Medieval Europe
Premodernist
Did LeBron James Play the 3 Greatest Games Ever in 2016?
Thinking Basketball
This Paradox Splits Smart People 50/50
Veritasium
The Banach-Tarski Paradox
Vsauce
The Real Reason We Don't Ride Zebras
CGP Grey
The Placebo Effect Has an Evil Twin
CGP Grey
Momentum isn't Magic: Hot Hand Analysis
Michael MacKelvie
A Better Way To Picture Atoms
minutephysics
Rewriting DNA to Make Custom Monstrosities
The Thought Emporium
One Minus One Plus One Minus One
Numberphile
Zeno's Paradox
Numberphile
Singapore: The World's Only Successful Dictatorship?
PolyMatter
Are America's Students Falling Behind?
PolyMatter
Anthropic Found Out Why AIs Go Insane
Two Minute Papers
Messages For The Future
Vsauce
Why Do India And China Have So Many People?
MinuteEarth
Reading
