learning from data

Benn Jordan investigates infrasound and low-frequency noise from data centers, including xAI's Colossus and a crypto mine, and reports his own experiments on the effects. The video covers the health and community impacts on nearby residents.

Scott Manley examines why SpaceX, Google and others are looking at orbital data centers. He walks through the economics, heat rejection and solar power arguments, with links to the Lumen/Starcloud and Google Suncatcher papers.

Scott Manley explains the thermal balance of spacecraft and how a satellite could be configured to radiate away the heat from datacenter-scale computing in space. His rough numbers suggest it is feasible.

Eager Space examines the practicality of putting data centers in orbit, walking through three scenarios with cost calculations and the launch cost reductions that would be required.

Scott Manley analyzes the idea of putting data centers in orbit, prompted by Lumen Orbit's funding. He examines the claimed advantage of constant solar power against launch cost, cooling and mass.

3Blue1Brown's first chapter on deep learning explains what neurons and layers are and the math underlying a basic neural network, using handwritten digit recognition with visual animations.

3Blue1Brown's visual explanation of how transformers and GPT-style language models work, tracing data through tokens, embeddings and attention.

Veritasium debunks the idea of learning styles, such as being a visual learner, with an interview with Daniel Willingham and a memory experiment. It argues that a multimodal approach helps everyone.

3Blue1Brown's visual walkthrough of the attention mechanism inside transformers, covering self-attention, multiple heads and cross-attention. It is chapter 6 of his deep learning series.

3Blue1Brown's third deep learning chapter gives a visual, intuitive account of backpropagation and what happens to a neural network as it learns. A follow-up video covers the underlying calculus.

3Blue1Brown's second neural network chapter explains cost functions and gradient descent, showing how a network learns by minimizing error during training.

Yosh trains a reinforcement learning agent to drive Trackmania until he can no longer beat it. The video explains how the training works and the design choices involved.

Veritasium explains that the Higgs boson accounts for only a small fraction of everyday mass. Most of a proton's mass comes from the energy of the gluon field binding its quarks.

Two Minute Papers covers OpenAI's multi-agent hide-and-seek experiment, in which agents learned emergent tool use and found ways to break the game's rules. It summarizes the 'Emergent Tool Use from Multi-Agent Interaction' paper.

Matt Parker of Stand-up Maths demonstrates MENACE, Donald Michie's 1961 machine made of matchboxes that learns to play noughts and crosses through reinforcement. He builds and trains a physical version.

3Blue1Brown's chapter 4 on deep learning gives a more formal, symbol-based account of backpropagation calculus, connecting the intuition from the previous chapter to the math found in textbooks and code.

3kliksphilip summarizes Valve's GDC talk on VACNet, a deep-learning system that flags CS:GO cheaters and feeds the Overwatch review queue. It covers Trust Factor, VAC, and how the data is collected and analyzed.

Hans Rosling uses augmented-reality animation to show 200 countries over 200 years in four minutes, built from 120,000 data points. It tracks life expectancy and income to show the world's development.

Veritasium examines which experts have real expertise, drawing on chess research and work by Klein and Kahneman. It explains that expertise comes from pattern recognition built through deliberate practice in valid, feedback-rich environments.

3Blue1Brown introduces Fourier series, from the heat equation to drawing arbitrary shapes with rotating circles. It builds a visual intuition for decomposing functions into sine waves.

A CAD-based animation from the Max Planck Institute for Plasma Physics showing how the Wendelstein 7-X stellarator is configured and what its fusion research aims to achieve. It runs about 11 minutes.

Linguist Simon Roper reconstructs how a London accent would have sounded from the 14th to the 21st centuries, reading passages in period pronunciation. It draws on the Cambridge History of the English Language.

Robert Miles explains reward modelling, in which an AI learns what humans want from their preference comparisons instead of a hand-written reward function. He walks through the 2017 paper on deep reinforcement learning from human preferences.

Kurzgesagt examines who is responsible for climate change, comparing historical, current and per-capita emissions by country. It argues that individual footprints are a distraction from larger responsibility.

CGP Grey explains jury nullification, a juror's power to acquit despite the law, and why mentioning it can get you struck from jury selection.

A short cartoon explaining how language models like ChatGPT work, from autocomplete and conditional probabilities to neural networks, gradient descent and backpropagation. It shows how networks act as universal function approximators.

A short excerpt from Frans de Waal's TED talk showing a capuchin monkey fairness experiment in which two monkeys are paid unequally, one with cucumber and one with grapes. The shortchanged monkey reacts angrily.

Alex O'Connor shows that ChatGPT cannot draw a wine glass filled to the brim, and ties the failure to David Hume's missing shade of blue. It asks what the failure says about how image models learn.

minutephysics, in collaboration with the US Census Bureau, explains differential privacy and why the 2020 census adds noise to protect individuals' data. It walks through how statistics can leak private information.

Engineering Explained argues that internal combustion engines are far from dead, and discusses the continued development of ICEs and the problems facing electric vehicles.

Musician Benn Jordan uses AI to detect AI-generated music and digs into music industry data, including streaming payouts and sample library copyright problems. He covers ways to tell whether a track is human-made.

A short video explaining the Intel Management Engine, a hidden embedded computer inside Intel machines with deep access to hardware. It is presented from an insider's perspective and raises backdoor and NSA concerns.

A short video telling how Xerox PARC invented the graphical user interface and how Steve Jobs's 1979 visit influenced Apple's Lisa.

Yuta asks whether native Japanese speakers can actually write kanji by hand, showing that many struggle to recall characters from memory.

A short animation from the DNA Learning Center showing how mRNA carries a gene's message to the ribosome, where tRNA and codons are used to assemble amino acids into a protein.

Paolo fromTOKYO explains why foreign tourists get arrested in Japan, covering strict laws and the justice system's very high conviction rate.

CGP Grey argues that your brain treats a whole home as one room during lockdown, blurring work, rest and leisure. He recommends creating separate spaces and routines for different activities to stay balanced.

Roy Beck's short gumball presentation using World Bank and Census data to argue that letting people immigrate to the US cannot solve world poverty because the numbers are so large. It is made by NumbersUSA, an immigration-restriction advocacy group.

A Kurzgesagt explainer on why the human population exploded and why growth is slowing, drawing on Our World in Data. It argues that birth rates fall as countries develop, easing overpopulation fears.

Guitarist Jim Lill methodically tests what in an electric guitar affects its tone, swapping parts on a partscaster and even trying a 2x4 body. He finds that wood and construction matter far less than people assume.