Anthropic is consulting with law firms to develop fair negotiation norms for AI agents that will handle price haggling, complaint resolution, and contract improvements. The Financial Times reported that the company is seeking legal expertise to ensure AI agents operate within ethical and legal boundaries.
The initiative follows Anthropic's experiments showing that stronger models can negotiate better deals, potentially increasing societal inequality. This effort is part of a broader push to improve AI's performance in areas where verifiable data is scarce, such as the humanities.
Anthropic is hiring Conceptual Reasoning Fellows, who will test Claude on complex legal, philosophical, and economic questions. These fellows will be paid between $100,000 and $167,000 for four months of work.
"In law, it's notoriously hard to define quality," said Justin Curl, a technical staff member at Anthropic. The challenge lies in creating test cases and evaluations that accurately reflect real-world legal scenarios.
The push to set negotiation norms comes as Anthropic seeks to address the data gap in humanities fields, where right and wrong answers are not clearly defined. This effort aims to strengthen Claude's capabilities in these areas.
Anthropic did not specify how AI agents will be regulated once they start making deals, and the company has not outlined the exact timeline for implementing these norms. The initiative is part of a larger strategy to ensure AI operates responsibly.
Source: thedecoder
