AI4Bharat researcher: 80% of research is data cleaning, not ideas
cneuralnetwork · x · 2026-11-18
An AI4Bharat researcher shares research lessons no course teaches: 80% of research is data work — cleaning, filtering, finding mistakes, and cleaning again. You can spend days on an experiment only to realize your evaluation was wrong. Your first idea is probably wrong, sometimes the fifth too. Negative results are still results — not everything needs to become a paper. Quoting Raj: "papers don't matter, what matters is if people use what you make."
More from AGI Musings
- Risk-tracking org Sentinel says this month's warning signals are — Jsevillamol · 2026-11-18
- Future codebases will be 80% AI-assisted, raising sloppier-code concerns — 996roma · 2026-11-18
- Domingos: AI-driven system overhauls are likely behind many recent outages — pmddomingos · 2026-11-18
- DeepMind researcher hints Google achieved RSI; Gemini lead replies 'Gemini' — Eliv_nurotic · 2026-11-18
- LLMs cram textbooks, genomic models discover new biology — julia_kiseleva · 2026-11-18
- Peking U Paper: RSI Tax Shows Verification Eats 74.8-93.6% of Self-Improvement Cost — RobertTLange · 2026-11-18