At a US military base in central California, autonomous all-terrain vehicles roam hillside trails, part of an initiative to train artificial intelligence (AI) models for future conflict zones. Founded by Coby Adcock and Collin Otis, Scout AI has secured $100 million to develop its 'Fury' model, aimed at operating military assets from logistics support to autonomous weapons.

The company's vision is to teach these AIs like soldiers, starting with foundational intelligence before honing their military-specific abilities. Their first test subjects are US Army vehicles during regular training cycles, with a focus on real-world scenarios that mimic unpredictable war zones.

Scout's approach to autonomy leverages Vision Language Action models (VLAs), recently introduced by Google DeepMind and now adopted by robotics start-ups like Figure.AI, led by Adcock’s brother. These models promise more general intelligence, making them adept at navigating complex terrains like those in training exercises.

While the technology is still experimental, it represents a significant step towards fully autonomous military operations. Critics might argue that as AI capabilities grow, so too does the potential for misuse. The question remains: can humanity maintain its commitment to peace amidst such rapid technological advancements?