Social Impacts of Artificial Intelligence and Mitigation Recommendations: An Exploratory Study
A structured taxonomy of AI social impacts drawn from 175 studies offers APS risk teams a consolidated reference for harm categorisation work.
Key points
- A 2023 systematic review of 175 articles identifies nine categories of AI social impact, led by bias and discrimination.
- The MIT AI Risk Repository has catalogued this as one of its ten foundational risk frameworks for comparative reference.
- This is a 2021 conference paper spotlighted in early 2025 - findings are well-established rather than novel.
Summary
MIT's AI Risk Repository has spotlighted a 2023 publication by Paes, Silveira, and Akkari that systematically reviewed 175 articles to identify nine categories of AI social impact, including bias and discrimination, risk of injury, data breach and privacy, job displacement, and environmental impacts. Bias and discrimination was the most frequently cited impact at 26%. The study also surfaces common mitigation strategies from the literature. As one of ten frameworks catalogued in the MIT AI Risk Repository, it forms part of a growing comparative collection of AI risk taxonomies that governance practitioners may draw on when structuring risk assessments.
Implications for Australian agencies
- Consider APS teams developing AI risk registers or harm taxonomies could consider cross-referencing this framework against existing Australian guidance such as the Responsible AI framework to check for coverage gaps.
- Monitor The MIT AI Risk Repository's broader collection of catalogued frameworks may be worth monitoring as a consolidated reference resource for comparative risk taxonomy work.
Implications are AI-generated. Starting points, not advice.
"Social Impacts of Artificial Intelligence and Mitigation Recommendations: An Exploratory Study" Source: MIT AI Risk Repository – Blog Published: 3 January 2025 URL: https://airisk.mit.edu/blog/social-impacts-of-artificial-intelligence-and-mitigation-recommendations-an-exploratory-study MIT's AI Risk Repository has spotlighted a 2023 publication by Paes, Silveira, and Akkari that systematically reviewed 175 articles to identify nine categories of AI social impact, including bias and discrimination, risk of injury, data breach and privacy, job displacement, and environmental impacts. Bias and discrimination was the most frequently cited impact at 26%. The study also surfaces common mitigation strategies from the literature. As one of ten frameworks catalogued in the MIT AI Risk Repository, it forms part of a growing comparative collection of AI risk taxonomies that governance practitioners may draw on when structuring risk assessments. Implications for Australian agencies: - [Consider] APS teams developing AI risk registers or harm taxonomies could consider cross-referencing this framework against existing Australian guidance such as the Responsible AI framework to check for coverage gaps. - [Monitor] The MIT AI Risk Repository's broader collection of catalogued frameworks may be worth monitoring as a consolidated reference resource for comparative risk taxonomy work. Retrieved from SIMS, 18 May 2026.