Urgent.News

What's breaking now, across thousands of outlets.

AI

ChatGPT-6 Astra plays World of Warcraft 'blind' and clears the orc starting zone in 40 minutes with no deaths

The model created its own World of Warcraft pathfinder on a private server with one prompt in OpenAI's Codex.

ChatGPT-6 Astra plays World of Warcraft 'blind' and clears the orc starting zone in 40 minutes with no deaths

OpenAI's GPT-6 Astra AI model demonstrated its capabilities by clearing the Orc starting area in World of Warcraft (WoW) within 40 minutes, with no deaths. The AI played the zone without visual input, relying solely on network traffic and quest data extracted from server files. Using a single prompt in Codex and an open-source client called agent-wow, the model navigated a private server to complete every quest in the Valley of Trials and reach Sen'jin Village.

Agent-wow, designed for autonomous AI agent players, relies on a module system for gameplay interactions, while AzerothCore, an open-source WoW server software, runs WoW 3.3.5a. The AI built 28 types of message modules to comprehend the game world, gathering quest data from AzerothCore's SQL files and pathfinding with a C++ helper and Detour library.

The agent's planning capabilities included completing prerequisite quests, selling junk, upgrading, and training abilities before entering the final cave. This WoW run marked a significant step for Astra, as it did not require screenshots or external resources, showcasing the model's raw potential in autonomous gameplay.

Written by urgent.news from Tom's Hardware's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at tomshardware.com →

More in AI

Your "Act As An Expert" Prompts Are Lying To You

Search "best ChatGPT prompts" and you'll get ten thousand lists, all free, all organized the same way: a category header, a block of italicized text starting with "Act as a world-class expert in X,"…

  • Prompt libraries lack crucial metadata that makes prompts effective.
  • Annotate prompts with specific job, scenarios, and failure modes.
  • Organize library by job, not persona, to guide prompt application.

More from Saturday 3 October →