Agentic Inequality

Sharp and coauthors distinguish three dimensions of agentic inequality: availability, quality, and quantity of agents. They argue that autonomous delegation and agent-to-agent competition can create asymmetries beyond earlier digital divides.^[1]

This is an analytical framework and research agenda, not a measured forecast of inevitable inequality. It connects Social Alignment with questions about who benefits from automation and Ghost Work. See Agentic Inequality and Agentic Inequality.

Multi-Agent Risks from Advanced AI supplies a complementary taxonomy of interaction failures; inequality in access and quality can shape the setting in which those interactions occur.

Bargaining performance

The Illusion of Rationality: Tacit Bias and Strategic Dominance in Frontier LLM Negotiation Games reports payoff asymmetries in controlled bargaining games. Such results motivate distributional evaluation without establishing the same effects in every market.

The Automated but Risky Game finds disparities in simulated consumer transactions. Model-estimated costs and negotiation prompts constrain how far those results can support claims about actual markets.