Please rotate your device to landscape mode to view the charts.

Data-Intensive Innovation and the State: Evidence from AI Firms in China

Journal: Review of Economic Studies

Date: 20230701

Author: Beraja, Martin; Yang, David Y.; Yuchtman, Noam

Abstract:
Developing artificial intelligence (AI) technology requires data. In many domains, government data far exceed in magnitude and scope data collected by the private sector, and AI firms often gain access to such data when providing services to the state. We argue that such access can stimulate commercial AI innovation in part because data and trained algorithms are shareable across government and commercial uses. We gather comprehensive information on firms and public security procurement contracts in China's facial recognition AI industry. We quantify the data accessible through contracts by measuring public security agencies' capacity to collect surveillance video. Using a triple-differences strategy, we find that data-rich contracts, compared to data-scarce ones, lead recipient firms to develop significantly and substantially more commercial AI software. Our analysis suggests a contribution of government data to the rise of China's facial recognition AI firms, and that states' data collection and provision policies could shape AI innovation.

Link: Google Scholar

Key Findings

Access to Government Data Drives Commercial AI Innovation

Firms receiving data-rich public security contracts produced 1.9 more commercial software products over 3 years compared to firms with data-scarce contracts - a 20.2% increase relative to pre-contract levels.

Dual Benefits from Government Contracts

Data-rich contracts led to increased production of both commercial and government software, overcoming potential resource crowding-out effects.

Data Sharing Drives Innovation

Government data and trained algorithms can be shared across government and commercial uses, enabling firms to develop AI products for broader commercial markets.

Software Production Before and After Data-Rich Contracts

  • Firms with data-rich contracts produced significantly more commercial software after contract receipt
  • No significant differences in pre-contract software production between firms
  • Effect grows over time, reaching 1.9 additional products after 3 years

Firm Characteristics by Contract Type

  • Data-rich contract recipients tend to be older firms
  • Higher capitalization among firms receiving data-rich contracts
  • More pre-contract software production by data-rich contract recipients

Growth in Chinese AI Development (2014-2018)

  • Rapid growth in China's share of global AI investment
  • Steady increase in AI contracts and software production
  • Parallel growth in surveillance infrastructure

Contribution and Implications

  • First causal evidence that government data access drives commercial AI innovation
  • Demonstrates how state surveillance capabilities can have unintended positive spillovers for commercial technology development
  • Suggests government data collection and provision policies could shape AI innovation across many sectors
  • Raises important questions about balancing innovation benefits against privacy and civil liberty concerns

Data Sources

  • Software production chart based on regression coefficients from Table 2
  • Firm characteristics comparison based on summary statistics from Table 1
  • Growth metrics based on Figure 1 showing relative changes from 2014 baseline