Core Insights - xAI is developing a project called "Macro Hard" aimed at utilizing idle computing resources from Tesla vehicles to train an AI agent capable of performing white-collar tasks, referred to as a "human simulator" [2][3] - The "human simulator" is designed to replicate human operations in digital environments, complementing Tesla's Optimus robot, which focuses on physical task automation [2] - The project features an efficient training mechanism leveraging xAI's supercomputing capabilities, allowing for daily model updates and minimizing traditional AI training issues [2] Group 1 - There are approximately 4 million Tesla vehicles in North America, most of which are idle and equipped with internet connectivity, cooling systems, and power sources [3] - The company plans to rent idle time from vehicle owners to run the "human simulator" on in-car computers, potentially saving significant costs [3] - The project has garnered significant attention from tech media, with some outlets praising its innovative approach to capital efficiency in AI agent architecture [3] Group 2 - The disclosure has sparked widespread discussion about the future of large-scale automated digital labor, with opinions divided on its potential to reshape the knowledge workforce and societal structures [3] - Concerns have been raised regarding potential risks associated with energy consumption, privacy protection, and antitrust scrutiny related to such projects [3] - Despite xAI's silence on the matter, the radical culture of the company may introduce legal and operational uncertainties [3]
马斯克秘密开发“人类模拟器”?要用百万辆特斯拉训练AI智能体