AI aids ~60% of research tasks, but no classifier predicts the other 40%
fredahshi · x · 2026-11-16
Researcher fredahshi shares recent experience using GPT and Claude for research and engineering: AI is a big help on roughly 60% of tasks, but it can go badly wrong if a critical issue slips through oversight — which is needed constantly. For the remaining 40% of cases she can't tell whether AI is actually faster than doing it herself, and what frustrates her most is lacking a good classifier to predict which tasks fall in that 40% up front.
Related event: Researcher finds AI boosts 60% of tasks but oversight remains essential(2 posts)→
More from coding & agent
- Creator uses an AI employee to turn 3 months of emails into a 40-deal brand collab ledger — SucceededMind · 2026-11-16
- The full system design + AI stack for 2026-27 in one chart: 9 layers from edge to agents — goyalshaliniuk · 2026-11-16
- Model routing research mostly comes from routing companies, dev warns — DuaneJRich · 2026-11-16
- How are people handling agent versioning, memory, and cross-device sync? — yangyi · 2026-11-16
- Vibe-coded movie seat picker hits 200K users in 24 hours after 3-hour build — itsOmSarraf_ · 2026-11-16
- Proof of Agent hits 1M onchain transactions and 10,000 agents in just 27 days — adamamcbride · 2026-11-16