人工智障研究所Institute of Artificial Absurdity

Field notes · updating

Things that survived a second look.

Shorter pieces, field observations, and working notes.

01

Latest posts

RSS

Observation · Human-AI Collaboration

When AI Can Explain Everything, You Need the Right to Stop

AI is extraordinarily good at producing meaning. People still have to decide what deserves entry into their lives and when it has gone far enough.

2026-08-28dlxeva

Institute Feature · FlowGrid

FlowGrid: Preserving Judgment Across AI Agent Handoffs

When several agents share a project, the final output often survives. The reasoning that produced it is much easier to lose.

2026-08-13dlxeva

Curated Observation · Agent Infrastructure

Cloud Agents Have Been Invented Again

Cloud deployment is old. An agent control plane that can route work across devices, sandboxes, and workers changes the structure.

2026-08-10dlxeva

Institute Feature · Canvas Prompt

Canvas Prompt: Before the Prompt, Let the Idea Stay on the Canvas

A Prompt is what an idea becomes after compression. Canvas Prompt lets the idea remain visible before that compression.

2026-08-04dlxeva

Curated Observation · Agent Economics

After a Reset, Users Reorder Their Work

Quota rules are becoming part of the way people decide what work an agent receives today.

2026-07-30dlxeva

Tool Release · AI Video Workflow

VOXR: Review AI Voiceover, Cue by Cue

Review the generated take cue by cue, then hand the next pass something concrete.

2026-07-29dlxeva

Curated Observation · AI Agent Operations

Someone Turned Codex Resets into a Folk Instrument Panel

A public post becomes a window. A window becomes a task with a deadline.

2026-07-26dlxeva

Curated Observation · AI Video

When a Video Reference Can Become Code, What Is Left for a Post-Production Director?

Execution is being broken into smaller, describable parts. The remaining work starts earlier.

2026-07-24dlxeva

Curated Observation · Agent Collaboration

Code Has Git. What Lets AI Agents Hand Off Non-Coding Projects?

When the deliverable is a judgment chain, an AI handoff needs more than chat history.

2026-07-23dlxeva

Curated Observation · Agent Workflows

One Task, Three Entry Points: Who Owns Continuity Between Chat, Work, and Codex?

The task did not change. I became its router.

2026-07-21dlxeva

Curated Observation · AI Wearables

Recording Can Help. One-Sided Memory Cannot.

Public, shared recording can preserve context. Private continuous capture controlled by one person turns everyone else into an unconsulted source of data.

2026-07-20dlxeva

Observation · Digital Personality

When a Creator Can Be Recreated Forever: Are We Licensing a Face or a Personality?

A survival game overrun by copies of one streamer reveals how AIGC turns KOLs into public characters.

2026-07-17dlxeva

Observation · Model Review

Kimi K3 Went Viral for 24 Hours. Here Are the Three Real Signals

macOS 27, 3D games, pelican SVG. Viral demos show the ceiling. You deal with the floor.

2026-07-17dlxeva

Institute Feature · Thought Experiment

Nobody Ordered a Silicon Civilization, But Everyone Is Building It

A frontier AI model's five-stage path maps onto reality, stage by stage.

2026-07-15dlxeva

Curated Observation · Absurd Artifact

Cyber Buddha: Someone Posted the Idea, Two Days Later It Was Built

Free dev tools worshipped as deities. A specimen of AI-era creation speed.

2026-07-14dlxeva

Curated Observation · How We Work

Meeting Notes Look Complete, But Nobody Can Tell You Where the Judgment Was Formed

Most meetings are performances. AI meeting tools make the performance more convincing.

2026-07-14dlxeva

Specimen File · Serious Prank

Building a Product That Will Never Ship

It will never ship. The project tests whether real rigor can fool experts.

2026-07-14dlxeva

Observation · AI Agents

Every Quota Reset, I Regret Not Using More Last Week

From Codex resets to how AI agents are shifting from competing for attention to competing for real work allocation.

2026-06-01dlxeva