Abstract
研究目的。本研究探讨人工智能（AI）市场预测工具如何影响钢铁企业的决策优化与经营绩效，重点分析其绩效效果、作用机制、实施障碍及数据驱动决策转型路径。
理论框架。本研究以数据驱动决策理论、资源基础观、动态能力理论以及技术–组织–环境（TOE）框架为理论基础。
研究方法。本研究采用混合研究方法。定量部分基于200家钢铁企业横向样本和45家企业纵向配对样本，运用描述性统计、独立样本T检验、多元回归和配对样本T检验分析吨材利润、流动资产周转率和存货周转次数等指标。定性部分基于两家典型钢铁企业10位管理者的半结构化访谈，并结合TOE框架进行主题分析。
研究发现与结论。研究发现，在样本期内，AI市场预测工具未对企业绩效产生显著的直接正向影响，但在提升信息质量、增强趋势识别和支持经营判断方面已发挥一定作用。研究同时表明，AI应用效果受到数据质量不足、系统集成不畅、数字化文化薄弱、复合型人才短缺以及行业高波动性等因素制约。总体来看，AI在钢铁企业中更多是一种决策支持工具，而非短期内直接提升绩效的工具，其价值具有明显的间接性、阶段性和情境依赖性。
建议。钢铁企业应将AI应用视为长期的数据驱动决策转型过程，重点加强数据基础建设、系统集成、人机协同决策机制、复合型人才培养以及高不确定环境下的弹性决策能力。
Purpose. This study examined how artificial intelligence (AI)-based market forecasting tools affect decision optimization and performance in steel enterprises. It focused on performance differences, decision mechanisms, implementation barriers, and pathways for data-driven decision transformation.Theoretical Framework. The study was grounded in data-driven decision-making theory, the resource-based view, dynamic capabilities, and the Technology–Organization–Environment (TOE) framework.
Methodology. A mixed-methods design was employed. Quantitative analysis used cross-sectional data from 200 steel enterprises and longitudinal matched data from 45 enterprises. Descriptive statistics, independent-samples t-tests, multiple regression, and paired-samples t-tests were applied to examine profit per ton of steel, current asset turnover, and inventory turnover. Qualitative analysis was based on semi-structured interviews with 10 managers from two representative steel enterprises and was analyzed through thematic analysis under the TOE framework.
Findings and Conclusion. The study found that AI market forecasting tools did not produce a significant direct positive effect on firm performance during the observed period. However, AI improved information quality, strengthened trend recognition, and supported managerial judgment. The findings further showed that implementation outcomes were constrained by weak data quality, insufficient system integration, limited digital culture, shortages of hybrid talent, and the high volatility of the steel industry. The study concluded that AI functions more as a decision-support tool than as an immediate performance-enhancing tool, and that its value is indirect, stage-dependent, and context-specific.
Recommendations. Steel enterprises should view AI adoption as a long-term decision transformation process rather than a short-term performance solution. They should strengthen data infrastructure, improve system integration, build human–machine collaborative decision mechanisms, cultivate hybrid talent, and enhance adaptive decision capacity in highly uncertain market environments.