deadly force

The first lecture of David Silver's reinforcement learning course, introducing the field and its basic concepts. The course follows the Sutton and Barto textbook.

Technology Connections explains what forced-air furnaces are, how they work and why they need electricity. He covers combustion, airflow and the parts of a modern furnace.

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.

A short clip from the Netflix series Space Force in which John Malkovich's character delivers a profanity-laden rant about Microsoft.

LegalEagle analyzes Katko v. Briney, the Iowa case about a homeowner whose shotgun booby trap injured a burglar, and the limits on deadly force in defending property.

A US Air Force film on the Heavy Press Program, which built some of the world's largest metal forging and extrusion presses for aircraft, starting from German presses taken after WWII.

A short smartphone-recorded clip from a 2014 event in Paris where DeepMind CEO Demis Hassabis presents an AI learning to play old Atari games. The system learns from only the video signal and the score. Audio quality is poor.

OpenAI's short demo of agents playing hide-and-seek in a simulated environment. Through self-play, the agents develop six progressively more complex strategies and counter-strategies, including tool use.

A two-minute Insider Tech clip on DeepMind's simulated agents that learned to walk, run, jump and climb without guidance, with the goofy resulting gaits.

Practical Engineering explains Mechanically Stabilized Earth, where layers of reinforcement make soil strong enough to hold heavy loads, shown with a sand structure supporting a car. It notes that MSE walls are everywhere in infrastructure.

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.

UC Berkeley EECS colloquium by OpenAI's John Schulman on reinforcement learning from human feedback. He covers why RL helps with truthfulness, hallucination, and the limits of behavior cloning.

Louis Rossmann argues that Apple designs and supports its products in ways that force planned obsolescence and obstruct independent repair. He draws on his experience repairing MacBooks and campaigns for right to repair.

A speculative-fiction video by Tom Scott, framed as coming from a future, about an AI called Earworm that enforces copyright so aggressively that humanity ends up forgetting a century of pop culture. It is animated and narrated as part of his series of videos from a future.

BigClive explains why microwave oven transformer projects are extremely dangerous, since the high voltage and high current can kill. He contrasts them with spot welder builds where the secondary is replaced by low-voltage heavy wire.

An IIHS news release video showing crash tests in which truck underride guards fail at relatively low speeds. The institute petitions for a stronger federal standard.

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.

A short clip of the McMurtry Speirling fan car setting a new Goodwood Hill record of 39.08 seconds at the 2022 Festival of Speed, driven by Max Chilton. It beat the previous timed shootout record set in 1999 and the outright record held by the VW ID.R.

Naomi Wu visits the Shenzhen offices of smartphone maker UMIDIGI in person to get them to release source code they owed under the GPLv2, and they comply.

A Computerphile video in which Rob Miles warns against assuming a general AI will think like a human. He uses the thought experiment of a stamp-collecting AI to show how a goal-driven system could pursue its aim with dangerous results.

Practical Engineering gives a short introduction to concrete, covering the difference between concrete and cement, the ingredients, and reinforcement. It includes a hydraulic press demonstration of compressive strength.

A short clip of Robert Sapolsky on how social hierarchy generates chronic stress, drawing on baboon research. It links rank-related stress to heart health and life expectancy.

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.

LockPickingLawyer shows a Sentry fire safe being cut in half with a saw in a few minutes. It is a short demonstration that these safes offer little protection against a determined thief.

Carwow's review of the McMurtry Spéirling, a 1,000hp electric fan car that generates huge downforce at standstill and reaches 0-60mph in about 1.4 seconds.

Not Just Bikes explains why crossing streets is dangerous in the US and Canada, where roads are designed to move cars fast, and contrasts it with Dutch designs for pedestrians.

Tom Scott tours the rotating room artificial gravity facility at Brandeis University's Graybiel lab. It shows how humans cope with rotation-based gravity and the odd effects such as Coriolis forces.

A short Australian dystopian film in which a man returns home to find citizens fining each other for infractions via a government app, with cameras everywhere.

Dwarkesh Patel interviews Richard Sutton, reinforcement learning pioneer, Turing Award winner and author of The Bitter Lesson. Sutton argues LLMs are a dead end because they cannot learn on the job, and that continual learning from experience will require a new architecture.

CNN report on bodycam footage of former Mesa police officer Philip Brailsford shooting unarmed Daniel Shaver in an Arizona hotel. Brailsford was later acquitted of murder charges.

Additional camera angles of the 1988 Sandia National Laboratories rocket-sled test in which an F-4 Phantom was crashed at about 500 mph into a block of reinforced concrete. The footage shows the aircraft disintegrating against the wall.

A short Unitree promo for its G1 humanoid robot, priced from $16K, showing agile motion, dexterous hands and imitation and reinforcement learning.

Graphic body-camera footage of the January 2016 fatal shooting of Daniel Shaver by an Arizona police officer in a hotel hallway, released after the officer's acquittal.

Footage of the US Navy's Phalanx CIWS, a radar-guided 20 mm Gatling gun, firing in live target practice. The Phalanx serves as terminal defense against anti-ship missiles.

A one-minute clip of David Mitchell on a UK panel show explaining why he respects poor customer service in Britain. He argues that people who do their job well should not also be required to be nice.

Dwarkesh Patel interviews Andrej Karpathy on why reinforcement learning is flawed, why AGI may blend into steady GDP growth, why self-driving took so long, and the future of education. The conversation runs about two and a half hours.

A Rocky Mountain IPv6 Taskforce talk in which T-Mobile's Stephan Lagerholm explains how the carrier went IPv6-only using DNS64 and 464XLAT, with a live Android demo and best practices.

Feature-length documentary following DeepMind's AlphaGo program as it prepares for and plays the 2016 match against Go champion Lee Sedol. It covers the team, the games, and the emotional reaction of the Go world to an AI beating a top human.

Ben Anderson's VICE documentary shows embedded reporting on Afghan security forces and US Marines in Helmand during the troop drawdown, exposing incompetence, drug abuse and corruption.

John Oliver's Last Week Tonight segment on civil asset forfeiture, the process by which police can seize property they suspect is tied to crime. It explains how the property itself is put on trial and how departments profit.