Scientific Idea Generation on AI Idea Bench 2025
0.75Reward NoveltyFlowPIE
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| FlowPIE2026.03 | 0.75 | 0.76 | 0.76 | 0.45 | 0.36 | 0.38 | 0.37 | 0.39 | |
| FlowPIEMethod variant=w/o isolation island2026.03 | 0.72 | 0.73 | 0.73 | — | — | — | — | — | |
| FlowPIEMethod variant=Initial Population2026.03 | 0.68 | 0.68 | 0.68 | — | — | — | — | — | |
| VirSci2026.03 | 0.6 | 0.65 | 0.63 | 0.27 | 0.17 | 0.2 | 0.2 | 0.21 | |
| Chain-of-Ideas2026.03 | 0.59 | 0.66 | 0.63 | 0.34 | 0.29 | 0.31 | 0.29 | 0.31 | |
| SCIPIP2026.03 | 0.5 | 0.73 | 0.62 | 0.38 | 0.37 | 0.3 | 0.29 | 0.34 | |
| Research Agent2026.03 | 0.49 | 0.66 | 0.58 | 0.22 | 0.19 | 0.13 | 0.13 | 0.17 |