Robust AI Evaluation through Maximal Lotteries

Hadi Khalaf, Serena Lutong Wang, Daniel Halpern, Itai Shapira, Flavio Calmon, Ariel D. Procaccia
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:56796-56827, 2026.

Abstract

The standard way to evaluate language models on subjective tasks is through pairwise comparisons: an annotator chooses the "better" of two model responses for a given prompt. These comparisons are then aggregated into a single ranking via the Bradley–Terry (BT) framework, forcing heterogeneous preferences into a total order and violating basic social-choice desiderata. In contrast, social choice theory provides an alternative approach called maximal lotteries, which aggregates pairwise preferences without imposing any assumptions on their structure. However, we show that maximal lotteries can be highly sensitive to heterogeneity among annotators and across prompts. We introduce robust lotteries, which optimize worst-case performance under plausible shifts in the preference data. On large-scale preference datasets, robust lotteries achieve more reliable win rate guarantees across the annotator distribution and recover a stable set of top performing models.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-khalaf26a, title = {Robust {AI} Evaluation through Maximal Lotteries}, author = {Khalaf, Hadi and Wang, Serena Lutong and Halpern, Daniel and Shapira, Itai and Calmon, Flavio and Procaccia, Ariel D.}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {56796--56827}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/khalaf26a/khalaf26a.pdf}, url = {https://proceedings.mlr.press/v306/khalaf26a.html}, abstract = {The standard way to evaluate language models on subjective tasks is through pairwise comparisons: an annotator chooses the "better" of two model responses for a given prompt. These comparisons are then aggregated into a single ranking via the Bradley–Terry (BT) framework, forcing heterogeneous preferences into a total order and violating basic social-choice desiderata. In contrast, social choice theory provides an alternative approach called maximal lotteries, which aggregates pairwise preferences without imposing any assumptions on their structure. However, we show that maximal lotteries can be highly sensitive to heterogeneity among annotators and across prompts. We introduce robust lotteries, which optimize worst-case performance under plausible shifts in the preference data. On large-scale preference datasets, robust lotteries achieve more reliable win rate guarantees across the annotator distribution and recover a stable set of top performing models.} }
Endnote
%0 Conference Paper %T Robust AI Evaluation through Maximal Lotteries %A Hadi Khalaf %A Serena Lutong Wang %A Daniel Halpern %A Itai Shapira %A Flavio Calmon %A Ariel D. Procaccia %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-khalaf26a %I PMLR %P 56796--56827 %U https://proceedings.mlr.press/v306/khalaf26a.html %V 306 %X The standard way to evaluate language models on subjective tasks is through pairwise comparisons: an annotator chooses the "better" of two model responses for a given prompt. These comparisons are then aggregated into a single ranking via the Bradley–Terry (BT) framework, forcing heterogeneous preferences into a total order and violating basic social-choice desiderata. In contrast, social choice theory provides an alternative approach called maximal lotteries, which aggregates pairwise preferences without imposing any assumptions on their structure. However, we show that maximal lotteries can be highly sensitive to heterogeneity among annotators and across prompts. We introduce robust lotteries, which optimize worst-case performance under plausible shifts in the preference data. On large-scale preference datasets, robust lotteries achieve more reliable win rate guarantees across the annotator distribution and recover a stable set of top performing models.
APA
Khalaf, H., Wang, S.L., Halpern, D., Shapira, I., Calmon, F. & Procaccia, A.D.. (2026). Robust AI Evaluation through Maximal Lotteries. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:56796-56827 Available from https://proceedings.mlr.press/v306/khalaf26a.html.

Related Material