AI signal, minus the noise.

This Medium article presents a poetic conversation reflecting on the concept of superposition in language models, the absence of subjective experience ('the cold a model never feels'), and the notion of meaning being worked out backwards. It offers a contemplative, non-technical exploration rather than concrete developments.

The article claims that artificial intelligence has not made programming obsolete, but rather has eliminated the human obligation to write code manually. It presents the viewpoint that while AI handles code generation, the broader practice of programming—problem-solving and system design—remains a human domain. The piece is a brief opinion without supporting data or concrete examples.

A designer shares a brief reflection that learning AI has not increased their design speed but is transforming their design identity and approach. The article provides no further concrete details, tools, or outcomes.

A hands-on tutorial streams 3,000 documents from the FineWeb sample-10BT subset without downloading the full multi-terabyte corpus. It reproduces quality filters (Gopher, C4, custom), finding most already-passed due to pre-filtering. MinHash-based deduplication with 128 permutations and 0.7 threshold identifies few near-duplicate pairs, consistent with per-crawl deduplication. GPT-2 token counts are verified against the stored field, showing near-perfect match (mean absolute difference ~0). Analytics cover token distribution, language scores, characters per token, and top domains, providing practical insights for scaling corpus preprocessing pipelines.