two-factor authentication

A WSJ report on China's 'dark factories', highly automated plants run by robots with little human presence, visiting Zeekr's EV factory. It covers Chinese labor costs, automation and overproduction in the EV industry.

Andrej Karpathy's first Neural Networks: Zero to Hero lecture, which builds micrograd, a tiny autograd engine, from scratch in Python. It explains backpropagation and neural network training step by step, assuming only basic Python and some high-school calculus.

Destin Sandlin tours United Launch Alliance's rocket factory with CEO Tory Bruno, showing friction stir welding, structural rocket skins and large milled parts.

Sentdex demonstrates GAN Theft Auto, a GameGAN-based neural network that recreates GTA V so the network itself is the playable environment. He shows the results and releases the code.

Geoffrey Hinton's 2025 Royal Institution Discourse on what AI has learned from biological intelligence and how it might surpass it. He explains neural networks and argues that human language use resembles what backpropagation-trained networks do.

Destin Sandlin of Smarter Every Day tours a Kodak factory to show how photographic film is manufactured. This is part one of a three-part series covering the large-scale coating and production process.

A very short clip from the TV series Westworld about whether it matters if you cannot tell a human from an artificial being.

A long Team House podcast interview with James Lawler, a former CIA operations officer who led counterproliferation work against the A.Q. Khan nuclear network. He discusses HUMINT, WMD and how such operations were run.

Engineer Stuff Made Here builds two new supposedly unpickable locks with different designs and sends them to the Lock Picking Lawyer to see whether they can be defeated. The video covers the novel anti-picking mechanisms.

Destin Sandlin tours a progressive metal stamping factory, showing machines that stamp millions of metal parts and the tooling behind them.

CGP Grey argues that the two-day weekend is poorly structured and proposes taking days off midweek, such as Wednesday and Sunday, instead.

Tom 7 (suckerpinch) delivers a long, comedic digression on gradient descent and floating-point imprecision. He shows how rounding behavior can supply the nonlinearity needed to build a neural network.

A FIDO Alliance demo with Google and Microsoft engineers showing passkeys working across operating systems, browsers, and devices. It walks through registering and signing in on new devices.

Part two of a StrictlyVC interview with OpenAI CEO Sam Altman, focused on AI. He discusses regulation, commodification of AI, Google's reluctance to release its own models, and best- and worst-case futures.

Sabine Hossenfelder explains how close nuclear fusion really is to break-even, and why headlines about ITER and lab milestones are misleading. She distinguishes plasma gain (Q plasma) from whole-system gain (Q total) and shows the latter is still far below one.

Veritasium explains why analog computing may suit AI workloads, covering old analog computers, The Analog Thing and Mythic's analog chips that use flash memory for neural network inference. It argues digital hardware is hitting limits.

3Blue1Brown explains why positive results on accurate medical tests are often false positives, and reframes Bayes' rule using likelihood ratios (Bayes factors) to update odds. It uses visual examples to make the reasoning intuitive.

BobbyBroccoli documentary on the rise of Nortel Networks, once the dominant Canadian telecom giant, during the dot-com era. It is part one of a two-part story.

Andrej Karpathy, then Tesla's head of AI, gives a CVPR 2021 workshop keynote on how Autopilot's neural networks work. A large part covers the move to a vision-only stack, including why radar was dropped and how vision produces depth estimates.

DIY Perks builds a webcam mod using a two-way mirror in a box so the camera sits behind the screen image. This allows real eye contact on video calls.

Jared Mauch tells how he got fed up with poor internet options in rural Michigan and built his own fiber network to his town. The talk was given at NLNOG Live in September 2020.

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.

Tesla's full AI Day 2022 presentation, covering the Tesla Bot hardware and software, Full Self-Driving planning and perception, the occupancy network, data labeling, simulation and Dojo training infrastructure.

Peter Shor tells the story of how he discovered Shor's Algorithm, from his early mathematics days to the quantum factoring breakthrough. It was recorded at MIT at the Physics of Computation conference.

Footage shot in New York City in 1911 by a Swedish film company, upscaled to 4K and 60 fps with neural networks and optionally colorized. The creator notes the colors are artificial and added only for ambiance.

Harvard professor Arthur Brooks explains why boredom is useful rather than something to avoid. He describes how a wandering mind engages the brain's default mode network, which supports creativity and curiosity, and suggests it may protect against depression.

GitHub interviews Linus Torvalds on twenty years of Git, from its creation in about ten days as a Linux kernel tool to its role in modern software development. He discusses technical decisions, community contributions and the project's philosophy.

Netflix engineer Drew Gallatin explains how Netflix serves TLS-encrypted video at 400Gb/s from a single FreeBSD server. He covers NUMA, kernel TLS, hardware TLS offload and the memory bandwidth and PCIe bottlenecks they hit.

Tech Tangents sets up original Doom's three-screen multi-monitor mode, which required a separate networked computer for each monitor. The video covers the hardware and setup work and the result.

Machine Learning Street Talk interviews Paul Lessard about Categorical Deep Learning, which uses category theory and monads to make neural network architectures more interpretable and composable. They also discuss neural networks' limits in reasoning and unbounded computation.

Andrej Karpathy, then Tesla's head of AI, presents at ScaledML 2020 on how Tesla builds full self-driving using vision-based neural nets and fleet data collection. He covers training custom detectors and neural nets gradually replacing traditional code.

Tom Scott-style absurdist research by tom7 for SIGBOVIK 2021, using machine learning to discover letterforms that are 'uppercaser' than uppercase A and lowercaser than lowercase. It is a comedic, technically inventive deep dive with fonts and a paper.

Dan Abramov's talk on the wet codebase, using visualizations to show how removing all duplication too early produces wrong abstractions that are hard to unwind.

Andrej Karpathy builds a bigram character-level language model from scratch, introducing torch.Tensor and the framework of training, sampling and loss evaluation. It is the first step toward a GPT-style Transformer in his makemore series.

John Ousterhout's Netdev 0x16 keynote arguing that nearly every aspect of TCP is wrong for the datacenter, from connections and streams to reliable delivery and congestion control. He presents the case for a replacement protocol.

A political commentary video arguing that tech billionaires such as Thiel, Musk and Andreessen, along with figures like Yarvin, are using the new administration to push their agenda. It connects Yarvin's ideas with Project 2025.

A Kurzgesagt video on the conditions in which most farm animals are raised. It argues that cruelty is for cheaper rather than better meat, and that a small price increase per meal could avoid most of the suffering.

Netflix engineer Drew Gallatin's EuroBSDcon 2022 talk quantifies the older FreeBSD optimizations that let Netflix serve web traffic at 800Gb/s, including TSO, LRO, VM page cache zones, asynchronous sendfile and unmapped mbufs.

A Welch Labs explainer on neural scaling laws and the compute-efficient frontier, asking why model loss seems to hit a line it cannot cross. It draws an analogy to an ideal gas law for AI.

A video presenting an Intel research paper on enhancing the photorealism of synthetic images. A convolutional network uses rendering intermediates, shown on GTA V footage that comes out looking like real street scenes.