xAI launches SpaceXAI persistent Grok Bot beta

The author argues that recent advances in AI agent capabilities—specifically their ability to operate computers across persistent cloud environments—make it rational to delegate all routine computer grunt work to them. Citing benchmarks like OSWorld and evaluations from METR, the author notes that current AI models now outperform humans on computer-operation tasks and handle software engineering tasks spanning many hours. The author contends that this shift frees human attention for higher-level work, though the outsized competitive advantage of doing so will diminish as adoption becomes widespread. Open article →