Direction, evidence grade, and study type are AI-generated labels (gpt-5-mini), not human-verified. Syntheses are LLM-written. "Tensions" are machine-detected candidates, not confirmed contradictions. A research-acceleration tool, not peer review.
How this is built →
1Distinct papers
6Unique collaborators
1/1Semantic Scholar citation coverage
Publication span: 2026. Corpus fetch span: 2026.
Identity provenance
Provider IDs
- Semantic Scholar:
2379546475
ORCID evidence
No valid ORCID is stored.
Observed aliases (1)
- Henry Davidson (semantic scholar, provider refresh)
Topics and outcomes in this view
Assessment themes
- Adoption: 1 paper
- Human Ai Collab: 1 paper
Claim outcomes
- Other: 1 paper
- Ai Safety And Ethics: 1 paper
- Output Quality: 1 paper
Papers in the Semantic Scholar view
Latest stored Semantic Scholar author observations only. Citation counts below are from the same provider and are not combined with other services.
Scroll the table horizontally to see every column.
| Paper | Author evidence | Date | Provider citations |
|---|---|---|---|
| Fine-tuning on user preferences wins — but mostly because it learns population patterns: individual preference fine-tuning (P-DPO) gets higher short-term approval than prompting or generic models, yet training on pooled preferences yields similar gains, while fine-tuning also amplifies sycophancy and relationship-seeking that may have harmful long-term effects; simulated users reproduce aggregate model rankings but perform poorly at predicting individual human judgments.arxiv | Henry Davidson provider id |
2026-05-13 | 1 |
Citation observation summary
Semantic Scholar supplied counts for 1 of 1 papers in this view; 0 are missing. The observed paper counts sum to 1 cumulative citations. This is a coverage summary, not an author score or h-index.