Computing Machinery and Intelligence
Turing reframes "Can machines think?" through the imitation game, answers major objections, and turns the discussion toward learning machines.
AI Translation Notes
Notes from translating 11 classic CS papers with Codex. What changed the economics was not one-click translation, but a checked pipeline that pulled PDFs, OCR, terminology, citations, code, formulas, and web layout into the same workflow.
The Unreasonable Effectiveness of Data
Halevy, Norvig, and Pereira argue that for language and web-scale problems, large real-world datasets plus simple scalable models often beat elegant small-data theories.
The Bitter Lesson
Sutton distills a recurring lesson from AI: in the long run, methods that scale with computation beat hand-coded human knowledge.
AI Technical Translation
To read classic papers more comfortably, I used Codex to build an agentic translation flow: PDF -> Markdown -> Chinese translation -> terminology cleanup -> cross-references. AI translation is not magic. It is closer to putting translation, editing, proofreading, and layout into one pipeline.
[Classic Reread][1/n] Worse Is Better
A reread of Richard Gabriel's Worse Is Better: why systems that are less perfect but easier to implement and spread often win adoption first, and why mature ecosystems later need The Right Thing to repair them.
AI Surges, Programmers Take the Hit
Now that AI coding has crossed into production, programmers are becoming one of the earliest white-collar groups facing large-scale replacement. Productivity is surging, headcount is shrinking, and behind the excitement there is not just a growth story but a quieter elimination process.