Informational Resilience Index (IRI) — Scores for 698 US Occupations, 2019-2024
收藏资源简介:
Companion dataset to the preprint "Informational Task Entropy and the Selection of Occupations Under AI-Driven Substitution" (Um 2026, Zenodo, DOI 10.5281/zenodo.20111476). This dataset provides Informational Resilience Index (IRI) scores for 698 US national-detailed cross-industry SOC occupations, calibrated on the BLS 2019-2024 panel. The IRI is defined as IRI_i = z(H_info_i) + z(JobZone_i) + z(log wage_i) with unit weights, where H_info = 1 - Eloundou alpha (informational task entropy proxy), JobZone is the O*NET preparation level (1-5), and wage is the BLS A_MEDIAN annual wage in 2024 USD. Each component is standardised. Validation: IRI correlates with 2019-2024 employment growth at Pearson r = +0.328 (p = 5.8e-19, N = 698). Robustness checks against the Felten AIOE alternative proxy and 2-digit NAICS industry fixed effects retain p < 1e-10. Caveats: IRI is a probabilistic indicator (R^2 ≈ 0.11), not deterministic prediction. International generalisation requires re-calibration. See companion preprint section 9.4 for full caveats. License: CC-BY 4.0. Replication, critique, and re-weighting welcome.
本数据集为预印本《信息任务熵与AI驱动替代下的职业选择》(Um,2026,Zenodo,DOI 10.5281/zenodo.20111476)的配套数据集。 本数据集包含698项美国全国细分跨行业标准职业分类(Standard Occupational Classification, SOC)职业的信息韧性指数(Informational Resilience Index, IRI)得分,该指数基于美国劳工统计局(Bureau of Labor Statistics, BLS)2019-2024面板数据校准。IRI的定义为: IRI_i = z(H_info_i) + z(JobZone_i) + z(log wage_i) 采用单位权重,其中H_info = 1 - 埃伦杜α(Eloundou α,信息任务熵代理指标),JobZone为美国职业信息网络(O*NET)的职业准备等级(1至5级),wage为美国劳工统计局发布的2024年美元计价的年薪中位数(A_MEDIAN)。各分量均经过标准化处理。 验证结果显示,IRI与2019-2024年就业增长的皮尔逊相关系数r = +0.328(p = 5.8×10^-19,样本量N = 698)。针对费尔滕AIOE替代代理指标以及2位码北美行业分类系统(North American Industry Classification System, NAICS)行业固定效应的稳健性检验结果均满足p < 1×10^-10。 注意事项:IRI仅为概率性指示指标(决定系数R² ≈ 0.11),并非确定性预测结果。若要进行国际推广应用,需重新校准指标。完整的注意事项详见配套预印本第9.4节。 许可协议:本数据集采用知识共享署名4.0(CC-BY 4.0)许可协议,欢迎对其进行复现、批评指正及重新加权。



