Preference optimization
Part of LLM alignment and post-training
Preference optimization is losing ground in AI research: 0.1% to 0.1% of new AI papers, -0.0 points. It ranks #171 of 214 by gain in share. As of Oct 10, 2026.
- Change in share
- -0.0 pts
- Share, last 3 months
- 0.1%
- Papers, last 3 months
- 36
- Ready for products
- 2.3 / 5
More in LLM alignment and post-training
Reinforcement learning post training0.2% → 0.2%+0.0 ptsSafety alignment0.3% → 0.3%+0.0 ptsHuman preference alignment0.0% → 0.0%0.0 ptsPreference alignment0.1% → 0.1%0.0 ptsEmergent misalignment0.1% → 0.0%0.0 ptsProcess reward model0.0% → 0.0%0.0 ptsProcess supervision0.0% → 0.0%0.0 ptsDirect preference optimization0.0% → 0.0%0.0 ptsPreference learning0.0% → 0.0%0.0 ptsReinforcement learning from AI feedback0.0% → 0.0%0.0 pts
Preference optimization: quick answers
No. Its share of new AI papers went from 0.1% to 0.1%, with 36 papers in the last 3 months.