Category: research
-
Pi Agent：Containerization 与 Compaction
TLDR: Coding agents need sandboxed execution, context compaction, and continuation mechanics so long-running work can survive safely across many tool calls.
-
Yuandong Tian Talks：搜索空间、RSI 与 Metaproductivity
TLDR: Search quality depends on shaping the action space, not only increasing rollouts; good representations make planning and learning much more effective.
-
Concordia：用 LLM Agent 搭建可干预的社会模拟
TLDR: Concordia treats LLM agents as situated social actors with memory, roles, and norms, making simulations easier to observe and intervene in than plain chat swarms.
-
Autocurricula and Multi-Agent Innovation：社会互动如何生成新问题
TLDR: Multi-agent intelligence should study how cooperation, competition, specialization, and shared discoveries create abilities that isolated agents would miss.
-
Social Dilemmas：三个经典社会困境
TLDR: Social dilemmas show why individually rational actions can damage group outcomes, and why cooperation depends on payoffs, repetition, reputation, and norms.
-
A Social Path to Human-Like AI：社会互动如何生成新数据
TLDR: Human-like AI may require populations of agents learning through social interaction, where cooperation and competition generate skills beyond single-agent training.
-
Talk with Shunyu Yao：AI 研究的反馈、系统与长期方向
TLDR: The conversation frames agent research as moving from raw model scaling toward long-horizon tool use, memory, personalization, science, and grounded reliability.
-
Anthropic Blogs：Harness Engineering 与 Context Engineering
TLDR: Long agentic tasks fail when context, tool use, and coordination drift; the useful lesson is to treat context engineering as runtime design.