How Replication Could Teach Machines What Good Science Looks Like — Edward Hughes

Building AI scientist systems requires shifting from goal-directed optimization to open-ended discovery, where agents proactively ask questions rather than merely answering them. Creativity functions by "satisficing"—respecting and occasionally breaking constraints—to navigate the maze of scientific knowledge. Edward Hughes, founder of Inherent, argues that true creativity involves meta-cognition to recognize the value of unexpected innovations. His Faraday agent demonstrates this by using a smaller model to instruct frontier coding agents in replicating research papers, proving that rigorous, forensic analysis of existing literature serves as a foundation for future discovery. This approach emphasizes a horizontal intelligence layer that leverages the complementarities between humans and AI, moving organizations toward a model of collective intelligence where agents and humans interact within a shared, evolving experimental ecosystem to accelerate scientific progress.