Anthropic is working with law firms to develop fair negotiation standards for AI agents capable of handling tasks like price haggling and contract improvements.
Key facts
- •Anthropic is consulting law firms to create fair negotiation norms for AI agents.
- •AI agents are being developed to handle price haggling, complaint resolution, and contract negotiations.
- •Conceptual Reasoning Fellows are being hired for four-month terms at salaries ranging from $100,000 to $167,000.
- •The program aims to improve Claude's reasoning in areas like law and philosophy where objective data is scarce.
- •Justin Curl, a member of Anthropic's technical staff, stated that defining quality in law is notoriously difficult.
Anthropic is consulting with law firms to establish fair negotiation norms for AI agents designed to act on behalf of users. These agents are expected to perform tasks such as resolving complaints, haggling over prices, and improving contract offers. The company's internal experiments indicate that more powerful models are capable of securing better deals, raising concerns about potential impacts on societal inequality.
Hiring Experts for Conceptual Reasoning
To improve Claude's performance in complex fields, Anthropic is hiring Conceptual Reasoning Fellows with backgrounds in law, philosophy, political science, economics, and cognitive science. These fellows will spend four months testing the model against judgment-heavy questions where clear-cut answers are often unavailable.
Addressing Data Challenges in Humanities
The initiative addresses a difficulty in training AI models for humanities-based tasks compared to math or coding, where verifiable data is more abundant. Justin Curl, a member of the technical staff at Anthropic, noted that defining quality in legal contexts is notoriously difficult, necessitating expert-led test cases and evaluations to strengthen the model.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by The Decoder.
