MIT's Ziming Liu: Evidence-Weighted Knowledge Base Lifts GPT-4o From 53% to 68%
ZimingLiu11 · x · 2026-10-24
In part 4 of a thread, MIT researcher Ziming Liu describes building an evidence-weighted knowledge base with Claude Opus 5, condensing up to 20 principles of experience. Without any weight updates, GPT-4o's accuracy on 50 held-out questions rises from 53.2% to 67.6%. Experience becomes transferable text—though Liu notes this remains incremental, not yet a science of AI.
More from Research
- A 17-state Beeping Busy Beaver machine that quasihalts iff the Collatz conjecture is false — rickasaurus · 2026-10-24
- NeurIPS paper OAT: attribute agent failures by learning only from success flows — SharonYixuanLi · 2026-10-24
- Redditor proposes crowdsourced multilingual agent benchmarks to escape English-only evals — Prestigiouspite · 2026-10-24
- Muon optimizer may nearly halve LLM training compute, challenging a decade of AdamW — bycloud · 2026-10-24
- MathDB launches to let mathematicians critique papers and AI-generated proofs — IgorCarron · 2026-10-24
- Knuth: 13th-century Sanskrit music scholars were doing computer science before computers — SumitGup · 2026-10-24