Large Language Models, and LLM-Based Agents, Should Be Used to Enhance the Digital Public Sphere

TL;DR

This paper advocates for using large language model-based recommenders to transform the digital public sphere by enhancing personalization, reducing data hoarding, and supporting autonomous attention management, addressing key technical challenges.

Contribution

It proposes a vision for LLM-based recommenders that prioritize democratic and human-centered values, outlining research hurdles and potential solutions.

Findings

Identifies four key research challenges for LLM recommenders.

Argues that overcoming these challenges can promote democratic digital spaces.

Synthesizes evidence of current harms with new LLM pipeline insights.

Abstract

This paper argues that large language model-based recommenders can displace today's attention-allocation machinery. LLM-based recommenders would ingest open-web content, infer a user's natural-language goals, and present information that matches their reflective preferences. Properly designed, they could deliver personalization without industrial-scale data hoarding, return control to individuals, optimize for genuine ends rather than click-through proxies, and support autonomous attention management. Synthesizing evidence of current systems' harms with recent work on LLM-driven pipelines, we identify four key research hurdles: generating candidates without centralized data, maintaining computational efficiency, modeling preferences robustly, and defending against prompt-injection. None looks prohibitive; surmounting them would steer the digital public sphere toward democratic, human-centered values.