Love for Believable AI: Artificial Partiality and Relationship Persistence as an Engineerable Stance
The study investigates creating artificial partiality in AI by introducing two components: caring (allocating interaction surplus above a baseline) and particularity (accumulating user-specific state estimates from history). The mechanism successfully reproduces the expected behavioral pattern but fails to match analytical magnitudes, suggesting it's a candidate for future perception studies rather than evidence of perceived love.