R0040/2026-03-28/Q002/S04
WebSearch — DPO for sycophancy reduction
Summary
| Field |
Value |
| Source/Database |
WebSearch |
| Query terms |
DPO sycophancy reduction Khan 2024 opinion sycophancy preference pairs |
| Filters |
None |
| Results returned |
10 |
| Results selected |
1 |
| Results rejected |
9 |
Selected Results
| Result |
Title |
URL |
Rationale |
| S04-R01 |
Mitigating Sycophancy in LLMs via DPO (Khan et al.) |
https://ieeexplore.ieee.org/document/10825538/ |
Primary source showing DPO reduces sycophancy 84-85% |
Rejected Results
| Result |
Title |
URL |
Rationale |
| S04-R02 |
Various results on DPO and sycophancy |
Multiple URLs |
9 results: 2 duplicates of previously captured papers, 2 tangential (social sycophancy, VLMs), 2 secondary analyses, 1 white paper, 1 student paper, 1 different topic (ICLR 2026 paper on internal origins) |
Notes
The Khan et al. paper was the primary target of this search, successfully located. The ICLR 2026 paper on "internal origins of sycophancy" is potentially relevant but was not accessible for detailed review.