Inscrutable AI?

For the entirety of human history, tools remained subordinate to people and their functioning was perfectly understandable to their makers. In a span of just a few years, Large Language Models have begun to speak, reason, persuade, and build interpersonal closeness more effectively than many people. Agentic systems now autonomously organize complex worklows, enable decisions at a pace and scale never achieved. As ageing populations drive labor shortages, societies are turning to these systems not out of simple economic advantage and convenience, but necessity.

AI systems are no longer simple tools. They have become entities, for which no current category adequately encompasses the abilities. This book confronts the ontological dilemma this shift has created. It examines the transformative cognitive and social abilities of current AI. It presents the accelerating deployment of LLMs and autonomous agents. It examines the risks that already accompany them. 

The book further proposes three foundations towards the production of governance frameworks: behavioral primacy and transparency-driven accountability, epistemic humility and scientific enquiry, and adaptive and proactive governance.

The decisions we make now will determine whether advanced AI remains a subordinate instrument of human flourishing or becomes something far more consequential.