Skip to main content
QUICK REVIEW

[论文解读] The Emerging Generative Artificial Intelligence Divide in the United States

Madeleine I. G. Daepp, Scott Counts|arXiv (Cornell University)|Apr 18, 2024
Ethics and Social Impacts of AI被引用 10
一句话总结

分析美国各县和各州在 ChatGPT 的知识/兴趣模式,基于 Bing 搜索数据,揭示教育驱动的数字鸿沟在生成式人工智能工具早期采用中的情况。

ABSTRACT

The digital divide refers to disparities in access to and use of digital tooling across social and economic groups. This divide can reinforce marginalization both at the individual level and at the level of places, because persistent economic advantages accrue to places where new technologies are adopted early. To what extent are emerging generative artificial intelligence (AI) tools subject to these social and spatial divides? We leverage a large-scale search query database to characterize U.S. residents' knowledge of a novel generative AI tool, ChatGPT, during its first six months of release. We identify hotspots of higher-than-expected search volumes for ChatGPT in coastal metropolitan areas, while coldspots are evident in the American South, Appalachia, and the Midwest. Nationwide, counties with the highest rates of search have proportionally more educated and more economically advantaged populations, as well as proportionally more technology and finance-sector jobs in comparison with other counties or with the national average. Observed associations with race/ethnicity and urbanicity are attenuated in fully adjusted hierarchical models, but education emerges as the strongest positive predictor of generative AI awareness. In the absence of intervention, early differences in uptake show a potential to reinforce existing spatial and socioeconomic divides.

研究动机与目标

  • 描述美国范围内对一种新型生成式人工智能工具(ChatGPT)的知识的空间分布模式。
  • 评估县级社会经济、人口统计与产业因素与对 ChatGPT 的兴趣之间的关系。
  • 在调整混杂变量后,检验教育是否是兴趣的最强预测因素。
  • 使用替代数据源(Google Trends)评估研究发现的稳健性。
  • 讨论降低新兴 AI 数字鸿沟的政策与设计含义。

提出的方法

  • 使用去标识化的 Bing 搜索查询数据(2022年12月至2023年5月),汇总到 ZIP、县和州层级,以衡量与 ChatGPT 相关的搜索。
  • 通过 HUD-USPS 对照表将 ZIP 码链接到县;对少于 50 个唯一用户的单元格进行抑制。
  • 计算州级和县级的 ChatGPT 搜索率,并使用 Moran’s I 与 Getis-Ord G* 统计量评估空间聚集性。
  • 使用 ACS 2016–2020 数据描述县级特征(教育、收入、农村性、种族/族群、产业结构)。
  • 拟合多层负二项回归模型,以总搜索量作为偏移量,预测 ChatGPT 搜索计数,包含州随机效应和连续的协变量区块。
  • 使用 Google Trends 数据和大都市区分析进行稳健性检验;检验残留空间自相关。
Figure 1: Rates of Search for ChatGPT by State. Colors indicate the fraction of all searches that included reference to ChatGPT in each state-period observation across the first six months since the initial public release of the generative artificial intelligence tool.
Figure 1: Rates of Search for ChatGPT by State. Colors indicate the fraction of all searches that included reference to ChatGPT in each state-period observation across the first six months since the initial public release of the generative artificial intelligence tool.

实验结果

研究问题

  • RQ1在发布后的前六个月内,美国各州和县对 ChatGPT 的兴趣的地理分布如何?
  • RQ2哪些县级社会经济、人口统计和产业因素与更高的 ChatGPT 搜索率相关?
  • RQ3在调整其他因素后,教育是否成为兴趣的最强预测因素?
  • RQ4观察到的模式是否对替代数据源和空间聚合层级具有稳健性?
  • RQ5这些发现如何与更广泛的数字鸿沟模式及政策含义相关?

主要发现

  • 对 ChatGPT 的兴趣呈现空间聚集特征,西海岸显示最高的搜索率,而阿巴拉契亚地区/海湾州则最低。
  • 搜索率较高的县往往更城市化、受教育程度更高、富裕、亚洲人口比例更大,并有更多科技/创意产业工作岗位。
  • 在完全调整的分层模型中,教育是对 ChatGPT 搜索率的最强正向预测因素,优于其他社会经济和人口因素。
  • 使用 Google Trends 的稳健性检验得到定性上相似的模式,支持主要结论。
  • 结果与已确立的二级数字鸿沟模式一致,表明在生成式 AI 工具的采用方面出现了新兴的不平等。
Figure 2: Average monthly rates of searches for ChatGPT by state. West coast states with the highest rates of search are highlighted in green; states with persistently low rates of search are highlighted in purple (Louisiana, Alabama, Mississippi) and blue (Tennessee, West Virginia, Kentucky).
Figure 2: Average monthly rates of searches for ChatGPT by state. West coast states with the highest rates of search are highlighted in green; states with persistently low rates of search are highlighted in purple (Louisiana, Alabama, Mississippi) and blue (Tennessee, West Virginia, Kentucky).

更好的研究,从现在开始

从论文设计到论文写作,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。