谷歌 Gemini 4 智能体无人类干预通关《我的世界》
Google Gemini 4 agents just beat Minecraft with no human intervention - Engadget
Engadget
以下为原文逐段中文翻译，保留图表与说明。AI 翻译并经 AI 对照复核，仍可能存在错误；核对数据和引用时请同时查看原文。译文发布：2026年10月19日 20:29（北京时间）。
智能体分工协作、互相预警威胁，甚至从阵亡队友那里取回物品。
图片来源：Google
谷歌 DeepMind 的“AlphaGo”人工智能因在围棋中获胜而闻名，但该公司最新的 Gemini 人工智能刚刚掌握了一款复杂得多的游戏。谷歌研究人员在一篇新论文中披露，四个 Gemini 4 Argon 智能体在无人类干预的情况下，各自独立行动，同时共享提示、威胁信息和错误教训，击败了《我的世界》的最终对手“末影龙”。这项实验表明，通用的智能体式人工智能模型如今已能通关复杂游戏，而这一成就过去需要定制的人工智能软件才能实现。
为攻克《我的世界》，Gemini 4 团队将四个 AI 智能体放入一个沙盒化、随机生成的《我的世界》生存世界，目标是击败末影龙。它们没有被分配角色、策略或针对该游戏的专用工具。每个智能体使用开源 Mineflayer 库控制一名玩家，并通过游戏内聊天和共享记忆与队友协作。
在游戏过程中，智能体很快自行决定分工。它们互相预警威胁、传授经验与技巧、提供坐标，甚至从阵亡队友那里取回物品。每个智能体都逐渐形成了自己的角色：例如，“janesick”制作工具，“gyldenia”收集食物和物资，“auratian”点亮关键传送门，“bulicame”搭建末影人庇护所并完成最后一击。
整个过程共耗时27.23小时，大部分时间花在末影龙上。“四个智能体编写并运行了2783个程序，交换了229条消息，途中死亡605次，”团队表示。第五个名为“the Observer”（观察者）的智能体作为独立见证者监控整个过程，既用于检查是否作弊，也用于发布实时报告和总结报告。这些智能体共使用了13.5亿个输入 token，最终成本按 Gemini 标价计算为810美元，按标准价格计算为1620美元。
“据我们所知，这是首次由 AI 智能体端到端自主通关《我的世界》，仅使用标准 Mineflayer 库，没有预先侦察的种子，也没有人类编写的其他游戏专用工具，”团队写道。“要获胜，智能体必须在超过一天的时间里，跨越数千个相互依赖的步骤保持连贯的计划。”
展开英文原文 · 与中文逐段对应
Google Gemini 4 agents just beat Minecraft with no human intervention - Engadget
Agents split up tasks, warned about threats and even recovered items from dead teammates.
Google's Deepmind "AlphaGo" AI is famous for winning at Go , but the company's latest Gemini AI just mastered a far more complex game. Four Gemini 4 Argon agents beat Minecraft's final " Ender Dragon" foe with no human intervention by working individually while sharing tips, threats and mistakes, Google researchers revealed in a new paper . The experiment shows how general agentic AI models can now beat complex games, a feat that used to required custom AI software.
To tackle Minecraft, the Gemini 4 team set four AI agents into a sandboxed, randomized Minecraft survival world with a goal to defeat the Ender Dragon. No roles, strategies or game-specific tools were assigned. Each agent controlled a player using the open-source Mineflayer library and coordinated with teammates through in-game chat and shared memory.
During gameplay, agents quickly elected to split up tasks. They warned each about threats, gave lessons and tips, provided coordinates and even recovered items from dead teammates. Each settled into roles: for example, "janesick" made tools, "gyldenia" gathered food and supplies, "auratian" lit a key portal and "bulicame" built Enderman shelters and landed the final blow.
All told, the process took 27.23 hours, with the majority of time spent on the Ender Dragon. "The four agents wrote and ran 2,783 programs, exchanged 229 messages and died 605 times along the way," the team said. A fifth agent called "the Observer" monitored the session as an independent witness, both to check for cheating and to publish live and summary reports. The agents used 1.35 billion input tokens and the final cost was $810 at Gemini's list price or $1,620 at the standard price.
"To our knowledge, this is the first end-to-end autonomous completion of Minecraft by AI agents, achieved using only the standard Mineflayer library without pre-scouted seeds or other game-specific tools written by humans," the team wrote. "Winning required the agents to keep a coherent plan across thousands of dependent steps for more than a day."
查看本期独立摘要
中文独立摘要 · AI 整理
这件事
谷歌 DeepMind 公布论文称，四个 Gemini 4 Argon 智能体在无人类干预下击败《我的世界》末影龙，全程27.23小时。智能体未预设角色、策略或游戏专用工具，仅通过游戏内聊天与共享记忆协作，自行分工造工具、收集食物、点亮传送门、搭建末影人掩体并完成最后一击。期间编写运行2783个程序、交换229条消息、死亡605次，消耗13.5亿输入 token，按标准价计成本1620美元。
本页概括所列来源的重点，属于独立摘要；完整报道请阅读原站。整理时间：2026年10月19日 20:31（北京时间）。
信息从哪里来
据谷歌 DeepMind 论文与 Engadget 报道，引述研究团队自述。
- Engadget 报道正文