top of page
  • Writer's pictureNicholas Kluge

AIRES at PUCRS Meta-analysis: an analysis on the global landscape of AI Ethics

Artificial Intelligence (AI) Ethics is an emerging field of research. However, this field has enough history and literature that its first meta-analyses have already been published.

In 2019, Anna Jobin, Marcello Ienca, and Effy Vayena mapped the countries responsible for producing the existing ethical guidelines for AI regulation, as well as other parameters (e.g., main ethical principles advocated). According to the authors, 84 documents were evaluated in this research.

Back in 2020, Thilo Hagendorff analyzed 21 of the most relevant documents to the international debate on AI Ethics published in the last five years, pointing out interesting blind spots in the field itself. At the same time, similar mappings have been conducted by institutions like the Future of Life Institute and the Berkman Klein Center (Harvard University).

According to the Future of Life Institute's initiative (Global AI Policy), only 34 countries are participating in this debate, which represents approximately 17% of all the countries that make up our "true global landscape".

Meanwhile, the report produced by the Berkman Klein Center, "Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-based Approaches to Principles for AI," points out that the AI Ethics debate converges around eight ethical principles, also pointing out how such principles are approached and interpreted by different global actors (i.e., governments, NGOs, private institutions, etc.).

In order to provide a more recent meta-analysis and with a higher level of granularity, we at AIRES at PUCRS will be publishing a meta-analysis of more than 200 papers published between 2014 and 2021.

Currently, we are in the process of post-processing the collected data, as well as analyzing our results. However, some preliminary results already allow us to project a good idea of the current landscape of Artificial Intelligence Ethics around the globe.

Such as the distribution of documents published by countries and international institutions:

More (and less) defended ethical principles:

Types of proposals for AI regulation:

Among several other parameters that were the focus of our analysis (e.g., the scope of impact, gender distribution among authors, types of institutions, types of documents, document size, the difference in the definition of ethical principles, etc.)

We hope to soon be sharing all our results in an interactive format accessible to everyone.

For more information, contact Nicholas Kluge (Lead Researcher).

16 views0 comments


bottom of page