ChatGPT-6 Astra plays World of Warcraft 'blind' and clears the orc starting zone in 40 minutes with no deaths
ChatGPT-6 Astra plays World of Warcraft 'blind' and clears the orc starting zone in 40 minutes with no deaths - AI agent navigates by parsing raw server network packets and SQL files OpenAI's model used the open-source agent-wow client to play on a private World of Warcraft server. OpenAI's GPT-6 Astra AI model was able to clear the Orc starting area in World of Warcraft (WoW) in 40 minutes with 0 deaths, agent-wow's developer says. The model played the zone without seeing a single rendered frame, relying on network traffic and quest data pulled from the server's own files. The approach used a single prompt in Codex with agent-wow, an open-source client, to play on a private server. The corresponding YouTube video says that the agent āstarts as a level 1 Orc, completes every quest in the Valley of Trials, and finishes the run in Senājin Village.ā Agent-wow is an āAzerothCore WoW client designed for autonomous AI agent players,ā according to the GitHub repo description. The client does not ādefine any gameplay mechanics such as movement, combat, or in-game interactionsā and instead only exposes a āmodule system for agents to buildā whatever they need to play. AzerothCore is open-source server software that runs WoW 3.3.5a, āthe final build of Wrath of the Lich King,ā and agent-wow talks to it over the gameās network protocol. The client does not connect to any live WoW server and instead can be run on local private servers for experimentation. The developer ran OpenAIās flagship model, released early last month, with extra high reasoning effort, one below the highest or maximum. The developer chose WoW for its mix of long-term strategy and short-term tactics, with an end goal of filling āan entire server with AI agents [to] see if they can clear Icecrown Citadel on heroic difficulty.ā The agent built a single module to grab 28 types of server messages, by our count, which are kept in memory. A Python script polls those messages to build the agentās picture of the world and sends messages back to act in it. The developer expected a higher-level implementation, but āIn practice, it was more than capable of working at the protocol layer.ā For quest data, the agent turned to data mining, pulling quest givers, turn-ins, and spawn points from AzerothCoreās SQL files. The developer says this is āarguably similar to how a human might spend hours on Wowhead to research quests,ā referring to a popular WoW info site. Going straight to the serverās own files arguably beats that, since theyāre the data the server itself runs on, while a fan siteās data may not always match. The projectās public workspace instructions also list AzerothCoreās source code as a resource for agents, though the developer doesnāt say whether this run used them. Some planning was used in the agentās approach. According to the developer, it worked through prerequisite quest chains in order, sold junk, equipped upgrades, and trained abilities before the zoneās final cave, and picked up both cave quests at once to complete them together. For pathfinding, the agent built a C++ helper that plots a route between a starting point and a destination with AzerothCoreās own navigation mesh files, or mmaps. The Detour pathfinding library finds the route, and the helper returns its waypoints as coordinates, or an error if thereās no complete path. The developer acknowledges that āfrom my research into heuristics-based bots, pathfinding is always one of the main challenges,ā but calls the agentās pathfinding abilities āoptimal.ā In fact, the agent was able to āexploit map bugsā at locations where collision properties may have been lacking. Get Tom's Hardware's best news and in-depth reviews, straight to your inbox. This is not the first game the model has tackled. Early last month, shortly after its release, it played the game Portal with screenshots and knowledge of the playerās position. This WoW run used no screenshots at all and only covered the first zone, while the Portal run went through the entire game over about 24 hours. Recently, developers have deployed various techniques to play PokĆ©mon Red, such as using a custom small model and a Jev decision model harness with Claude coaching. Next up for the developer: discovering whether a single agent can reach level 80 completely on its own, and whether multiple agents can team up to complete content through the gameās social features. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds. Shane Downing is a Contributing Writer for Tomās Hardware, covering consumer storage, PC hardware, and AI.
Comments
No comments yet. Start the discussion.