AI-Driven Agents with Prompts Designed for High Agreeableness Increase the Likelihood of Being Mistaken for a Human in the Turing Test
원문 초록
Large Language Models based on transformer algorithms have revolutionized Artificial Intelligence by enabling verbal interaction with machines akin to human conversation. These AI agents have surpassed the Turing Test, achieving confusion rates up to 50%. However, challenges persist, especially with the advent of robots and the need to humanize machines for improved Human-AI collaboration. In this experiment, three GPT agents with varying levels of agreeableness (disagreeable, neutral, agreeable) based on the Big Five Inventory were tested in a Turing Test. All exceeded a 50% confusion rate, with the highly agreeable AI agent surpassing 60%. This agent was also recognized as exhibiting the most human-like traits. Various explanations in the literature address why these GPT agents were perceived as human, including psychological frameworks for understanding anthropomorphism. These findings highlight the importance of personality engineering as an emerging discipline in artificial intelligence, calling for collaboration with psychology to develop ergonomic psychological models that enhance system adaptability in collaborative activities.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Hallucination Mitigation using Agentic AI Natural Language-Based Frameworks
RS-Agent: 지능형 에이전트를 통한 원격 탐사 작업 자동화
RS-Agent: Automating Remote Sensing Tasks through Intelligent Agent
최근 멀티모달 대규모 언어 모델(MLLM)의 발전은 시각적 질의응답 및 장면 이해와 같은 원격 탐사 작업에서 가능성을 보여주었다. 그러나 기존 모델들은 기본적인 인스트럭션 팔로잉에 머물러 있으며, 다중 소스 데이터 통합, 세밀한 공간 추론, 도메인 전문 지식을 요구하는 실세계 시나리오에서는 한계를 보인다. 이러한 격차를 해소하기 위해, 우리는 구조화된 작업 계획 및 도구 오케스트레이션을 통해 사용자 의도를 전문적인 원격 탐사 워크플로우와 연결하는 도메인 적응형 지능 에이전트 RS-Agent를 제안한다. RS-Agent는 전형적인 원격 탐사 워크플로우에 맞춰 네 가지 구성요소로 이루어진다: 의도 이해 및 프로세스 계획을 위한 중앙 컨트롤러(Central Controller), 도구 실행을 위한 동적 툴킷, 작업별 전문가 가이던스를 위한 솔루션 공간(Solution Space), 그리고 도메인 지식 지원을 위한 지식 공간(Knowledge Space)이다. 또한 작업 유형을 식별하고 전문가 정의 솔루션을 검색함으로써 계획 수립을 향상시키는 Task-Aware Retrieval과, 검색된 지식의 관련성과 완전성을 높이는 가중 이중 경로 검색-증강 생성(RAG) 방법인 DualRAG를 새롭게 도입한다. RS-Agent는 광학 영상과 SAR(합성 개구 레이더) 영상을 포함한 다양한 촬영 모달리티를 기본 지원하며, SAR 전용 처리 도구를 실행 가능한 워크플로우로 자동 구성할 수 있다. 9개 데이터셋과 18개 원격 탐사 작업에 대한 실험 결과, RS-Agent는 최신 MLLM 대비 현저히 높은 성능을 달성하였으며, 95% 이상의 작업 계획 정확도와 장면 분류, 객체 계수, 원격 탐사 시각적 질의응답에서 우수한 결과를 보였다. 이 결과는 지능형 지리공간 분석을 위해 LLM 추론과 원격 탐사 전문성을 결합하는 것의 가치를 입증한다.
$τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
Comparative Analysis of AI Agent Architectures for Entity Relationship Classification
ThinkTank: A Framework for Generalizing Domain-Specific AI Agent Systems into Universal Collaborative Intelligence Platforms
AgentCPM-GUI: Building Mobile-Use Agents with Reinforcement Fine-Tuning
토론
이 논문과 연결된 포럼 글이 아직 없습니다.