Genomics researcher: Claude and Codex always fail on assay design questions
anshulkundaje · x · 2026-09-26
A shared discussion points out that certain question types trip up Claude and Codex regardless of model or reasoning depth. Genomic assay development is one example: the models lack knowledge of library structure, which ends to sequence, or where enzymes act — unless the user repeatedly corrects them.
The takeaway: frontier models have systematic knowledge blind spots tied to domain-specific tacit knowledge, which stronger reasoning alone cannot fix.
More from Models
- Observation: Astra uses filler tokens far more effectively than other models — scaling01 · 2026-09-26
- Ethan Mollick: 'Keep prompts short' is bad advice, and minimizing token cost confuses inputs with outputs — emollick · 2026-09-26
- GPT-6 Luna uses fewer reasoning tokens than 5.6 on ARC-AGI-2, hard tasks stymie both — mhmazur · 2026-09-26
- Heavy user: fast, cheap Claude Opus 5.5 now takes all my serious work — brandon_galang · 2026-09-26
- Model excels at Sokoban-style puzzles, sparking questions about training data contamination — lukaszkaiser · 2026-09-26
- GPT-6 Astra's First Draft Fooled Every's CEO: Big Writing Upgrade, But Some Bad Habits — every · 2026-09-26