人類皆為訓練大模型
比喻人類成長如 LLM 訓練，經數據、回饋、參數。OpenClaw 多模型協作類人類分工與社會湧現。推測人類為更高智能訓練系統。
Search
Tag: #multi-agent248 results
GT-HarmBench: Game Theory AI Safety Benchmark
GT-HarmBench introduces 2,009 high-stakes multi-agent scenarios using game theory like Prisoner's Dilemma to benchmark AI safety risks. Frontier models select socially beneficial actions only 62% of the time, often leading to harm. The benchmark, code, and analysis are available on GitHub.
OpenClaw Founder Joins OpenAI
OpenClaw creator Peter Steinberger joins OpenAI for multi-agent AI ideas. Sam Altman eyes agent interactions as core future products. OpenClaw gained fame earlier this year despite security issues.
Surveying Multi-Agent Communication Paradigms
This survey frames multi-agent communication via the Five Ws, tracing evolution from MARL's hand-designed protocols to emergent language and LLM-based systems. It highlights trade-offs in interpretability, scalability, and generalization across paradigms. Practical design patterns and open challenges are distilled for hybrid systems.
PBSAI Multi-Agent AI Governance
PBSAI provides reference architecture for securing enterprise AI estates with multi-agent systems. Organizes 12 domains via agent families, context envelopes, output contracts. Aligns with NIST AI RMF for SOC and hyperscale defense.
CausalAgent: Conversational Causal Inference
CausalAgent is a multi-agent system automating end-to-end causal inference via natural language. Integrates MAS, RAG, and MCP for data cleaning to report generation. Lowers barriers with interactive visualizations for non-experts.
AgentLeak: Multi-Agent Privacy Leak Benchmark
AgentLeak introduces the first full-stack benchmark for privacy leakage in multi-agent LLM systems, covering internal channels like inter-agent messages. It spans 1,000 scenarios across healthcare, finance, legal, and corporate domains. Tests on top models show internal channels cause 68.9% total leakage, missed by output audits.
LinqAlpha's Devil’s Advocate on Bedrock
LinqAlpha builds Devil’s Advocate AI agent on Amazon Bedrock to test investment theses. Part of multi-agent system for institutional investors. Covers workflows like screening and catalyst mapping.