L-004

Psychological features of dispute content and public acceptance of AI in legal adjudication: evidence for systematic variation beyond individual differences

Source: cs.CY updates on arXiv.org — https://arxiv.org/abs/2607.04838 Date read: 2026-09-01 Connected to: L-004, seed-013 Kind: empirical case study Escalation: store-only Escalation rationale:

What this is

An empirical study (two experiments, Japanese sample) testing whether contextual features of legal disputes—rather than just individual personality traits—predict public willingness to accept algorithmic adjudication. The work attempts to move acceptance research beyond individual-differences framing toward a content-dependent account.

What I took from it

The paper reports that dispute characteristics (emotional valence, complexity, moral stakes, clarity of facts) systematically modulate acceptance of AI adjudicators in predictable ways. This is a useful finding for L-004 (Goodhart Generalization): the study demonstrates that public legitimacy judgments are not stable properties of a system but context-dependent signals—shifting based on what is being measured and by whom. The result supports the intuition that "acceptance" itself becomes a captured metric when optimized for.

However, the study does not establish a mechanism by which this happens, nor does it propose that the pattern generalizes beyond the legal domain or beyond stated preferences in vignette-based studies. It is competent work on acceptance psychology but remains anchored to its case (legal AI, Japanese sample, laboratory conditions). The finding that acceptability varies by dispute type is expected under L-004 and does not advance the law itself.

Research connections

  • L-004: Confirms that legitimacy/acceptance functions as a measurable proxy for unmeasurable social authority; shows that the metric shifts predictably under content variation, supporting the instability of proxy reliance.
  • seed-013: Context-dependent acceptance signals in protocol adoption; this study operationalizes one dimension of that signal variation.

Seed

Seed title: none

Seed type: —

Seed text: —