TRACES grades the process, not the answer: six-dimension eval for open-ended AI science
Faheem_uh · x · 2026-09-11
TRACES (built by ApodexAI) evaluates AI on real, open-ended, genuinely unresolved problems where no answer key exists. Instead of checking final answers, it scores six dimensions:
- T (Tools): picks and uses the right instruments correctly
- R (Repair): catches and fixes its own mistakes
- A (Alternatives): holds competing hypotheses instead of committing too early
- C (Coherence): stays consistent across a long investigation
- E (Evidence): grounds every claim in real, traceable data
- S (Scope): states where the conclusion does and doesn't hold
The analogy: a new drug can take years to prove itself — reality itself hasn't delivered the answer yet, so answer-key-style benchmarks fall short.
Related event: Apodex unveils TRACES benchmark for discovery-oriented AI(3 posts)→
More from Research
- ParetoRSI-Gym benchmarks whether AI agents can recursively self-improve LLMs across domains — Haojin Wang · 2026-10-28
- Stanford's Ψ-0.6 is a unified vision-language world model with adaptive reasoning streams — Stanford · 2026-10-28
- Shanghai AI Lab's OpenESM shows energy-steered models can scale as foundation learners — Shanghai-AI-Laboratory · 2026-10-28
- CellTrack-Bench tests VLMs on microscopic cell tracking, exposing reasoning gaps — biohub · 2026-10-28
- Hybrid LLM agent scores 83.2%, beats nine VLA policies on new SO-101 robot arm benchmark — Truman Hickok · 2026-10-28
- DeepMind researcher: all of mathematics will soon be formalizable, with agents hunting proofs — ChrSzegedy · 2026-10-28