agent harness

Mario Zechner explains why he built pi, a minimal coding agent, out of frustration with existing agent harnesses. He also discusses how AI agents are straining open source projects and argues for slowing down.

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 pieced-together clip of the Agent Smith interrogation scene from The Matrix (1999), where Smith delivers his monologue about humanity.

A short clip from The Matrix in which Agent Smith tells Morpheus that humans behave like a virus. It is shared as a standalone monologue.

A recording from real hardware of Tim Follin's multichannel 1-bit music for Agent X on the ZX Spectrum. It shows several voices coaxed out of the single-bit beeper.

Hyperland is a 1990 BBC documentary written by and starring Douglas Adams, who tours an imagined future of hypertext and software agents. It features Ted Nelson among others.

A long interview with former KGB agent Yuri Bezmenov describing what he calls ideological subversion, a process of demoralizing a nation. He claims it works by influencing academia and media over years.

Apple's late-1980s Knowledge Navigator concept video, produced under John Sculley, showing a professor using a tablet-like device with a conversational software agent. It envisions hypertext, touchscreens and voice assistants.

Steve Yegge shows Efrit, a tool that gives Claude full control of Emacs, on the Vibe Coding with Steve and Gene show. Claude opens info pages, runs shells and even draws a Mandelbrot fractal in a buffer.

Ron Garret tells the story of the Remote Agent Experiment on NASA's Deep Space 1, where autonomous software controlled a spacecraft in 1999. He describes debugging Lisp code remotely from millions of miles away.

A 1984 interview with former KGB agent Yuri Bezmenov on ideological subversion and psychological warfare, reposted in 2020 with a modern political framing.

A re-upload of American Airlines captain Warren Vanderburgh's 1997 presentation warning that pilots who over-rely on cockpit automation lose manual flying skills and awareness. The uploader adds context from his own flying.

Rich Hickey's long presentation on Clojure and concurrency covers locking and immutability, then goes in depth on refs, transactions and agents. It ends with a multithreaded ant colony simulation demo.

Jon Gjengset spends about three hours livecoding complex changes in a Rust codebase with agentic AI tools, testing his hypothesis that AI is weak at non-pattern work. He documents where Claude Code helps and where it struggles.

Steve Mould messes with an automated phone bot by telling it to forget its previous prompts and give him a bolognese recipe. It is a short demo of prompt injection against an AI voice agent.

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.

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

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.

Sebastian Lague explores an ant-inspired algorithm and simulates ant and slime mold behavior, using compute shaders. Source code for both simulations is available.

Primer simulates green beard altruism, where individuals recognize and help others carrying a shared marker, and shows how imposters affect its evolution. It is an animated evolution simulation based on published papers.

Andrej Karpathy's keynote at AI Startup School laying out software 1.0, 2.0 and 3.0, with LLMs seen as utilities, fabs and operating systems. It covers LLM quirks, partial-autonomy apps and keeping a human in the generate-and-verify loop.

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.

Internet of Bugs analyzes Cognition's Devin demo frame by frame, arguing the video did not show Devin completing paid Upwork jobs as claimed, and showing what the task required versus what Devin produced.

A short video of David Foster Wallace's 'This Is Water' commencement speech about awareness and choosing how to think. It argues that the most obvious realities are the hardest to see, using the fish-in-water parable.

Robert Miles explains the orthogonality thesis: how an agent can be highly intelligent while pursuing goals that seem stupid or alien to humans. He separates intelligence from goals and touches on the stamp-collector example.

Victoria Warmerdam's Oscar-winning Dutch short film in which a woman fails a series of Captcha tests and is plunged into a strange new reality. It is a dark, Black Mirror-like sci-fi comedy.

Footage of British steeplejack Fred Dibnah climbing a chimney overhang at over fifty years old, without a harness. It is a short clip from his steeplejack work.

Rob Miles explains on Computerphile why implementing an off switch for a general AI is hard. A capable agent would resist being stopped because shutdown prevents it from achieving its goal.

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.

Demis Hassabis of DeepMind speaks at Google Zeitgeist, saying human-level AI is decades away but the debate should start now. He covers DeepMind's focus on game-playing agents.

The keynote launch video for the Rabbit r1 handheld AI device, presenting its assistant and a teachable mode for agents. It is a product announcement with live-style demos.

Stripe's John Collison interviews Anthropic CEO Dario Amodei about growth to around $5 billion ARR, model economics, agentic AI and vitalism.

Apple's 1987 Knowledge Navigator concept video imagines a conversational voice assistant, multi-touch tablet and video calling used by a professor. It was made years before the graphical web existed.

A short clip from the film Spies Like Us in which unprepared characters take a CIA agent test and Chevy Chase cheats his way through it.

A short hyperlapse of a 2019 Toyota Corolla using comma.ai's OpenPilot 0.6.3 to drive 100 miles with no intervention except for the windshield wipers.

Associated Press footage of masked federal agents detaining Tufts student Rumeysa Ozturk on a street in Somerville, Massachusetts. Her lawyer filed an emergency motion seeking her release.