Work

Human-powered AI Gym: Lessons Learned as the Test and Evaluation Team for the DARPA SHADE Program: Human-powered AI Gym

Year: 2024

Type: conference-paper

Abstract: We present the best practices and collaborative methods used during the DARPA SHADE program, exploring the use of AI agents to advise diplomatic negotiations. In this accelerated 18-month program, we created an environment for seven research performers to collaborate in the development of an AI Gym. This AI Gym provided an area in which AI developed by the various teams could compete, negotiate, and explain their intentions as they played the game of Diplomacy, allowing us to evaluate their abilities against other digital and human agents.

Source: Practice and Experience in Advanced Research Computing

DOI: 10.1145/3626203.3670579

Authors: Justin Drake, Niall Gaffney, Nathan Freeman, Erik Ferlanti, John Fonner

Institutions: Texas Advanced Computing Center, The University of Texas at Austin

Language: English

Topic: Artificial Intelligence in Games

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funder: Universitas Brawijaya

OpenAlex ID: W4400799124

FWCI: 0.153

Cites: 6

Cited by: 1

Related to: 10