aiqbench analysis
This research challenges the standard reliance on local pairwise comparisons for AI alignment and participatory design, demonstrating that they often fail when users hold multiple global priorities or experience internal conflict between them. By formally modeling these pluralistic preferences, the study shows that allowing indecision significantly improves preference learning efficiency while avoiding behavioral distortions caused by forced choices. Ultimately, this work advocates for new elicitation methods that directly capture complex value structures rather than forcing binary local judgments.
Internal Pluralism and the Limits of Pairwise Comparisons
arXiv:2607.02672v2 Announce Type: replace Abstract: Local pairwise comparisons are a standard tool for learning how people want decision rules to work, e.g., in participatory design or alignment. However, their use builds in two strong assumptions: that local comparisons are sufficient evidence about how a person wants an automated decision rule to behave, and that people can always answer those comparisons decisively. We investigate how these assumptions may be compromised under internal pluralism: the idea that an individual evaluates decision rules according to multiple authoritative priorities about how the rule should behave. We provide a formal model of such pluralistic preferences over decision rules, which then lets us identify two distinct failures of forced local pairwise comparison data. First, priorities such as proportionality, egalitarianism, and equal treatment are inherently global: what they imply in one case can depend on what happens elsewhere, so local comparisons may fail to capture them. Second, even when priorities are representable locally, tension between strongly-held priorities can generate internal conflict, producing potentially costly behavioral distortions when comparisons are forced. We then use our model to investigate the alternative -- allowing people to report indecision -- and our findings suggest that doing so can considerably reduce the number of queries needed to learn preferences accurately. We conclude by describing how our model points toward preference-learning methods that elicit these priorities directly, yielding more faithful and interpretable accounts of what people value.
Abstract:Local pairwise comparisons are a standard tool for learning how people want decision rules to work, e.g., in participatory design or alignment. However, their use builds in two strong assumptions: that local comparisons are sufficient evidence about how a person wants an automated decision rule to behave, and that people can always answer those comparisons decisively. We investigate how these assumptions may be compromised under internal pluralism: the idea that an individual evaluates decision rules according to multiple authoritative priorities about how the rule should behave. We provide a formal model of such pluralistic preferences over decision rules, which then lets us identify two distinct failures of forced local pairwise comparison data. First, priorities such as proportionality, egalitarianism, and equal treatment are inherently global: what they imply in one case can depend on what happens elsewhere, so local comparisons may fail to capture them. Second, even when priorities are representable locally, tension between strongly-held priorities can generate internal conflict, producing potentially costly behavioral distortions when comparisons are forced. We then use our model to investigate the alternative -- allowing people to report indecision -- and our findings suggest that doing so can considerably reduce the number of queries needed to learn preferences accurately. We conclude by describing how our model points toward preference-learning methods that elicit these priorities directly, yielding more faithful and interpretable accounts of what people value.
Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
arXiv:2607.02672 [cs.AI]
(or arXiv:2607.02672v2 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2607.02672 Focus to learn more arXiv-issued DOI via DataCite
https://doi.org/10.1145/3821539.3827910 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 335 words
Related coverage
Hint-based CoT faithfulness evals still mostly work on Claude
LessWrong · Jul 30, 2026
Does ChatGPT really have a strong left-wing bias?
LessWrong · Jul 27, 2026
Enhancing LLM Metacognition via Cognitive Pairwise Training
arXiv · Aug 25, 2026
AMD Instinct MI455X vs MI355X: A Technical Look at the Advancing AI 2026 Inference Numbers
AMD · Oct 8, 2026