What are you optimizing for? Aligning Recommender Systems with Human Values

TL;DR

This paper examines how recommender systems can be aligned with human values like fairness and diversity, highlighting current practices and proposing future directions involving stakeholder participation and interactive learning.

Contribution

It identifies the limitations of current value engineering in recommender systems and suggests four key directions inspired by AI alignment research for better stakeholder involvement and complex value learning.

Findings

Current practices address issues one at a time with limited stakeholder involvement.

Values engineering relies on classifiers from human-labeled data.

Proposes four directions: measures of alignment, participatory design, interactive learning, and informed judgments.

Abstract

We describe cases where real recommender systems were modified in the service of various human values such as diversity, fairness, well-being, time well spent, and factual accuracy. From this we identify the current practice of values engineering: the creation of classifiers from human-created data with value-based labels. This has worked in practice for a variety of issues, but problems are addressed one at a time, and users and other stakeholders have seldom been involved. Instead, we look to AI alignment work for approaches that could learn complex values directly from stakeholders, and identify four major directions: useful measures of alignment, participatory design and operation, interactive value learning, and informed deliberative judgments.