Letting AI Formulate the Research Question: Autonomous Research Agents
Source check: Source checked. Not independent scientific confirmation.
Finding: An AI-agent-run analysis compares annual AlphaEarth/TESSERA reference scores with one-year CONUS tree-cover labels; reference/control corrections did not consistently improve screening, while the supervised classifier scored higher with unequal label resources. The 24 gain positives and partition-sensitive contrasts keep the finding narrow.
Technology: The study analyzes learned AlphaEarth/TESSERA annual embeddings and derived scores; the author repository records that AI agents formulated, conducted and wrote the study
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- The study analyzes learned AlphaEarth/TESSERA annual embeddings and derived scores; the author repository records that AI agents formulated, conducted and wrote the study.
- Test and result
- An AI-agent-run analysis compares annual AlphaEarth/TESSERA reference scores with one-year CONUS tree-cover labels; reference/control corrections did not consistently improve screening, while the supervised classifier scored higher with unequal label resources. The 24 gain positives and partition-sensitive contrasts keep the finding narrow. Evaluation: The source evaluates supplied initial classes on held-out one-year LCMAP pairs (109 gains, 337 losses; 24 gain positives) and a separate Annual NLCD Forest test. It reports unequal label resources for the supervised comparator, reversal of all eight correction contrasts under exploratory partition changes, and limitations against multi-year restoration or ecological-recovery claims.
- Limitation / still unknown
- The result is limited to the reported one-year CONUS screening comparison; partition sensitivity and unequal label resources prevent a multi-year restoration or ecological-recovery claim.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://raw.githubusercontent.com/kofujimura/autonomous-research-agents/1501e89c40d0e700a9d9e7042f6442f8c93e5c5e/runs/run-014/paper.md
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.