When Generative AI Devalues Quantity Signals: Strategic Reallocation on Online Programming Platforms
Paper Type
Short
Paper Number
PACIS2026-1349
Description
Information asymmetry in software recruitment drives employers to rely on Online Programming Platforms (OPPs), where submission counts traditionally served as costly, reliable signals. However, Generative AI (GAI) drastically lowers coding costs, inflating quantity metrics and triggering "signal devaluation." How do users maintain distinctiveness amidst these devalued signals? Applying Synthetic Difference-in-Differences to a large panel of HackerRank users around ChatGPT's release, we identify a "strategic reallocation": users shift their effort from submission volume toward task difficulty, focusing on complex problems where AI still struggles. Notably, low-skill users escalate difficulty to capture signal premiums, whereas high-skill incumbents strategically withdraw to avoid information noise. Crucially, we isolate this signaling motive from pure productivity gains from GAI, demonstrating that only "signal-vulnerable" users lacking external credentials drive this escalation. This study provides critical insights for reshaping talent screening and platform governance in the AI era.
Recommended Citation
Ru, Zeyuan; Tang, Heng; and Wang, Qian, "When Generative AI Devalues Quantity Signals: Strategic Reallocation on Online Programming Platforms" (2026). PACIS 2026 Proceedings. 17.
https://aisel.aisnet.org/pacis2026/ai_fow/ai_fow/17
When Generative AI Devalues Quantity Signals: Strategic Reallocation on Online Programming Platforms
Information asymmetry in software recruitment drives employers to rely on Online Programming Platforms (OPPs), where submission counts traditionally served as costly, reliable signals. However, Generative AI (GAI) drastically lowers coding costs, inflating quantity metrics and triggering "signal devaluation." How do users maintain distinctiveness amidst these devalued signals? Applying Synthetic Difference-in-Differences to a large panel of HackerRank users around ChatGPT's release, we identify a "strategic reallocation": users shift their effort from submission volume toward task difficulty, focusing on complex problems where AI still struggles. Notably, low-skill users escalate difficulty to capture signal premiums, whereas high-skill incumbents strategically withdraw to avoid information noise. Crucially, we isolate this signaling motive from pure productivity gains from GAI, demonstrating that only "signal-vulnerable" users lacking external credentials drive this escalation. This study provides critical insights for reshaping talent screening and platform governance in the AI era.
Comments
02-FutureofWork