Overall breakdown of AI in the World presentations
Topics covered a roughly even split between corporate, public, and internet/digital society AI use.
While presentations highlighted a diversity of AI use cases by companies, fairness concerns predominantly centered around disparate impact.
Industry clusters where AI fairness issues have been raised
AI use cases where AI fairness issues have been raised
Breakdown of AI fairness issues identified
Relationship between industry, AI use cases and fairness issues
Corporate America
In Corporate America, the potential for AI to cause disparate impact across social groups is by far the most pressing fairness issue.
lorem ipsum
Public Sphere
In the public sphere, disparate impact across social groups is the most frequently raised fairness issue. Optimizing human tasks and processes is particularly complex and high-stakes.
Internet Society
Recommendation systems are the most common use of AI in the internet society. In the same vein, epistemic injustice is the most often raised fairness issue.
Here, feedback loops are especially tight and crucial, both to the success of the software companies, and leads to potentially harmful feedback loops for the users.
Relationship between AI use cases and fairness issues
Complex relationship between how AI is used and fairness issues raised.
For optimization algorithms, classification, NLP, the most pressing fairness concern was disparate impact among social groups.
For LLM & chatbots and recommendation systems, the most pressing fairness concern was epistemic injustice (all types).