machine language

Rob Miles explains on Computerphile how language models generate plausible text and how transformers improved on recurrent networks, with a look at what comes next.

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.

Johnny Harris investigates why McDonald's ice cream machines are so often broken, covering the Taylor machines, the repair restrictions on franchisees and the Kytch workaround device.

Technology Connections takes apart the step-by-step electromechanical logic of a 1970s pinball machine. He shows how relays, steppers and switches track state and score without electronics.

Rob Miles discusses GPT-3 on Computerphile, covering how scaling parameters kept improving results over GPT-2 and how the model nearly learned arithmetic without anyone knowing exactly how.

Technology Connections takes a long, detailed look at how a 1970s electromechanical pinball machine works, from relays to stepper motors and mechanical logic. It is the first part of a series.

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.

Tom Scott, with linguists Gretchen McCulloch and Molly Ruhl, examines whether the language you speak changes how you think. The short answer is mostly no, with caveats about the weak version of the Sapir-Whorf hypothesis.

Veritasium short showing Drew Berry's scientifically accurate animations of the molecular machines that drive cell division. It walks through mitosis, chromosomes and motor proteins.

An illustrated 23-minute summary of David Friedman's talk based on The Machinery of Freedom. It describes how law, police and courts could work as market services in a stateless society, covering private law, polycentric law and objections.

Branch Education's animated deep dive into ASML's EUV photolithography machine and how it prints nanometer-scale transistors onto chips. It covers the system's components and the process of chip manufacturing.

Jared Owen's 3D animated breakdown of how a Brunswick GS-X bowling pinsetter works, showing its mechanisms for sweeping, lifting and resetting pins and balls.

A Business Insider World Wide Waste episode on H&M's garment recycling machines, which turn used clothing into new apparel. It asks whether a fast fashion company can solve the waste problem it helped create.

Rob Miles explains "glitch" tokens, odd strings in large language model vocabularies that make the model produce gibberish. He covers how they likely arise from tokenizer training data, such as Reddit usernames, that never appeared in the model's own training.

Steve Mould explains motion amplification and video magnification, techniques that find tiny changes in video invisible to the eye and exaggerate them. He covers the MIT research and industrial uses such as diagnosing vibrating machinery.

Ray Dalio's 30-minute animated video explains how the economy works through credit, debt, interest rates, and short- and long-term cycles. It is based on his template for understanding the economy.

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

Episode 4 of Mike Maloney's Hidden Secrets of Money, explaining how currency is created through a chain of IOUs among the Treasury, banks and the Federal Reserve. It covers bonds, Fed checks, debt, taxes and fractional reserve lending, from a gold-and-silver advocate's viewpoint.

A short Al Jazeera animation narrated by Amy Goodman presenting Noam Chomsky's five filters of the mass media: ownership, advertising, media elite sourcing, flak and a common enemy. It is part of the MediaTheorised series.

Vox explains the Berlin and Kay research showing that languages around the world tend to invent color words in the same order. It walks through the hierarchy and the debate around it.

Veritasium explores compliant mechanisms, devices that move by flexing rather than using hinges and joints, with researchers at BYU. It covers advantages such as fewer parts and precision, including uses in nuclear weapon safety.

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

Michael Wesch's five-minute video about Web 2.0, showing how the read/write web, XML and user-generated content change how people create and share information.

Tom Scott makes a short descriptivist argument that language keeps changing and that casual internet writing conventions, like lowercase typing and dropped full stops, are legitimate. It is part of his Language Files series, written with Gretchen McCulloch.

A short Harvard Online animation, narrated by Professor Rob Lue, showing how ATP synthase works inside mitochondria. It comes from the HarvardX Cell Biology: Mitochondria course.

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

Mike Pound explains the discrete cosine transform and how it underpins JPEG compression. It is part two of Computerphile's JPEG series.

Huygens Optics explains the Fourier transform and shows how a lens performs one optically. A DLP projector is used as a Fourier filter to do simple pattern recognition on lithographic character patterns.

Benn Jordan explains how Flock Safety's license plate cameras log and share vehicle locations, and explores how adversarial AI can defeat them. It covers weak security and the surveillance risks of private police-camera networks.

A Computerphile video in which Dr Mike Pound and PhD student Lewis Stuart explain 3D Gaussian splatting, a quick technique that turns photos of a scene into a 3D model without much compute. It includes a demo and compares the approach to NeRFs.

FortNine looks at why Tesla Autopilot has rear-ended motorcyclists, speculating that the vision system misreads motorcycle taillights as a distant car. It argues that radar and other sensors would help.

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.

Veritasium traces analog computers from the Antikythera mechanism and Kelvin's tide-predicting machines to fire control and modern analog approaches. It explains how these machines computed continuously rather than digitally.

The Map Men explain why British place names are so hard to pronounce, tracing it to layers of Viking, Anglo-Saxon, Celtic and Norman influence, with a pronunciation guide.

Aaron Francis walks through how PHP has changed from 2012 to 2023, covering traits, array destructuring, generators, typed properties and more. It is a quick chaptered tour of modern PHP features.

A Veritasium video on AlphaFold, DeepMind's protein structure prediction system, and the protein design work it enabled. It features interviews with John Jumper and with David Baker's protein design group and explains why predicting protein folding matters.

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.

Computerphile video in which Mike Pound explains a University of Maryland paper on watermarking LLM output. The method biases token choices so generated text can later be detected statistically.

Asianometry explains the EUV light source inside ASML's lithography machines, where lasers hit falling tin droplets twice to produce 13.5 nm light. It walks through the engineering of the droplet generator, the pre-pulse and main laser, and the plasma that results.

An animated introduction to geometric algebra, presented as a language that can describe most of physics. It was made with manim and covers multivectors and how they generalize familiar tools like cross products and quaternions.