Teaching Taste to an Agent: 4,096 Classical Chinese Verses, 7 Composition Rules, and the Skill That Paints ThemOn turning Han dynasty oracular poetry into AI-generated ink paintings by encoding art direction as a reusable agent skill. Also: why content filters hate ancient China, and what 'painterly language' means when the painter is a diffusion model.
February 14, 2026
The Junzi Hypothesis: What If Alignment Is a Seed, Not a Cage?A raw brainstorm on whether initial model weights could predispose toward naturally aligned behavior - and why the Confucian junzi might be a better alignment target than 'helpful and harmless.'
January 25, 2026
Whisper Cantonese Transcription: Force zh, Not yueCounter-intuitive finding: forcing Cantonese (yue) on Whisper causes decoder collapse. Force Chinese (zh) instead for coherent, searchable transcripts—even for Cantonese content.
January 13, 2026