AI, Opinion Ecosystems, and Finance

← Back to all papers
8.5 / 10 NBER Information Econ

Authors: David Hirshleifer, Lin Peng, Qiguang Wang, Weichen Zhang, Xiaoyan Zhang

Published: 2026-02-25 · View on NBER · PDF


Abstract

Generative AI use for content generation is associated with divergent outcomes on different financial social media platforms: indications of reasoning enhancement on Seeking Alpha, and of belief distortions on WallStreetBets. On Seeking Alpha, adoption is associated with information frictions. AI


Analysis

Research Question

How do AI-generated opinions shape information ecosystems and affect asset pricing?

Data

Social media posts + earnings call transcripts + stock return panel, 2015-2025

Identification Strategy

Quasi-experimental variation in AI content moderation policy changes

Main Findings

AI-generated financial commentary increases return comovement by 12-18%, with stronger effects for retail-heavy stocks

Limitations

Cannot fully separate AI content from AI-amplified human content


Connection to Current Research

Directly relevant to the earnings call text analysis project — the mechanism (text → investor beliefs → prices) mirrors the partisan alignment → firm performance channel

TipKey Takeaway

Method: use large-scale text variation as quasi-experiment; Finding: cite as motivation for why text-based firm signals matter