Anthropic is currently collaborating with law firms to establish ethical standards for AI agents capable of conducting negotiations. The initiative aims to define norms for automated systems that will handle tasks such as price bargaining, complaint resolution, and contract optimization.
Internal experiments conducted by the company indicate that more advanced models achieve superior negotiation outcomes. Experts express concern that this capability could potentially exacerbate existing societal inequalities if left unregulated.
To address these challenges, the company is recruiting Conceptual Reasoning Fellows with legal backgrounds. These specialists will spend four months evaluating the performance of the Claude model on complex tasks involving law, economics, and political science.
The compensation for these fellowship positions ranges from $100 000 to $167 000. The program focuses on creating robust test cases, as the company struggles with a lack of verifiable data in humanities compared to mathematics and coding.
