Sane General Intelligence: A Technorealist Frame for AGI with Hugo Latapie
Details
Join us for this online event as we welcome back Hugo Latapie, Founder of Taijitu AI and formerly a leader of AI/AGI research at Cisco. He will be discussing his recent research.
AGI is usually discussed as a future capability threshold: a machine system able to reason, learn, plan, and act across domains. But even before that threshold is reached, the conversation needs a second question: what would make any generally capable intelligence worth trusting?
This talk introduces Sane General Intelligence as a proposed category for grounded, corrigible, bounded, and reality-attuned general intelligence, and presents technorealism as a pro-technology frame that separates capability from authority. Rather than assuming that more intelligence automatically routes to better outcomes, the talk asks what conditions are needed for powerful AI systems to support truth, agency, correction, and human flourishing.
Sane General Intelligence: A Taxonomy for Grounded and Bounded General Intelligence (Preprint)