Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Kantian-Utilitarian XAI: Meta-Explained

Zahra Atf, Peter R. Lewis

TL;DR

The paper tackles ethical decision-making in consumer contexts by integrating Kantian deontic checks with utilitarian welfare scoring in a gamified XAI. It deploys dual symbolic engines (Kantian and utilitarian) plus a meta-explainer with a regret bound of $0.2$ to switch to deontically clean near-parity options when welfare loss is small. Results across eight synthetic coffee scenarios show that the combined+meta explanation preserves near-utilitarian welfare while substantially reducing deontic violations, and half of Kantian–utilitarian conflicts are resolved via the regret mechanism. The work provides auditable configurations and policy traces, demonstrating tunable alignment between norm enforcement and welfare in actionable consumer guidance and highlighting avenues for human studies and domain expansion.

Abstract

We present a gamified explainable AI (XAI) system for ethically aware consumer decision-making in the coffee domain. Each session comprises six rounds with three options per round. Two symbolic engines provide real-time reasons: a Kantian module flags rule violations (e.g., child labor, deforestation risk without shade certification, opaque supply chains, unsafe decaf), and a utilitarian module scores options via multi-criteria aggregation over normalized attributes (price, carbon, water, transparency, farmer income share, taste/freshness, packaging, convenience). A meta-explainer with a regret bound (0.2) highlights Kantian--utilitarian (mis)alignment and switches to a deontically clean, near-parity option when welfare loss is small. We release a structured configuration (attribute schema, certification map, weights, rule set), a policy trace for auditability, and an interactive UI.

Kantian-Utilitarian XAI: Meta-Explained

TL;DR

The paper tackles ethical decision-making in consumer contexts by integrating Kantian deontic checks with utilitarian welfare scoring in a gamified XAI. It deploys dual symbolic engines (Kantian and utilitarian) plus a meta-explainer with a regret bound of to switch to deontically clean near-parity options when welfare loss is small. Results across eight synthetic coffee scenarios show that the combined+meta explanation preserves near-utilitarian welfare while substantially reducing deontic violations, and half of Kantian–utilitarian conflicts are resolved via the regret mechanism. The work provides auditable configurations and policy traces, demonstrating tunable alignment between norm enforcement and welfare in actionable consumer guidance and highlighting avenues for human studies and domain expansion.

Abstract

We present a gamified explainable AI (XAI) system for ethically aware consumer decision-making in the coffee domain. Each session comprises six rounds with three options per round. Two symbolic engines provide real-time reasons: a Kantian module flags rule violations (e.g., child labor, deforestation risk without shade certification, opaque supply chains, unsafe decaf), and a utilitarian module scores options via multi-criteria aggregation over normalized attributes (price, carbon, water, transparency, farmer income share, taste/freshness, packaging, convenience). A meta-explainer with a regret bound (0.2) highlights Kantian--utilitarian (mis)alignment and switches to a deontically clean, near-parity option when welfare loss is small. We release a structured configuration (attribute schema, certification map, weights, rule set), a policy trace for auditability, and an interactive UI.

Paper Structure

This paper contains 3 sections, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Research Methodology
  3. Results, Discussion & Outlook

Figures (2)

  • Figure 1: Condition-level effects on (a) welfare uplift and (b) violation-free choices across six synthetic coffee rounds. The combined + meta condition preserves near-utilitarian welfare while improving deontic compliance.
  • Figure 2: One-column architecture: synthetic coffee stimuli and ethical KB feed dual symbolic engines (Kantian + utilitarian); a meta-explainer enforces regret-bounded switches and generates rationales for the gamified UI. All runs are logged for exact reproducibility.