A Playbook for Securing AI: What I Learned About Safety, Control, and the Fight for a Good Future
What happens when someone pulls the wrong levers? A continuation of the AI Triad, this piece works through misuse vs. misalignment, the three-layer defensive playbook, and why d/acc might be the most coherent philosophy for navigating this transition.
Read EssayRecent Essays
Latest Dispatches
A Playbook for Securing AI: What I Learned About Safety, Control, and the Fight for a Good Future
What happens when someone pulls the wrong levers? A continuation of the AI Triad, this piece works through misuse vs. misalignment, the three-layer defensive playbook, and why d/acc might be the most coherent philosophy for navigating this transition.
The AI Triad: What I Learned About AI and National Security from My BlueDot Course
Three things make modern AI possible: algorithms, data, and computing power. Understanding how they interact is becoming one of the most important things a policymaker can know.
How Models Actually Make Predictions
Part 2: Loss functions, optimization, and how systems improve themselves step by step.
A Playbook for Securing AI: What I Learned About Safety, Control, and the Fight for a Good Future
What happens when someone pulls the wrong levers? A continuation of the AI Triad, this piece works through misuse vs. misalignment, the three-layer defensive playbook, and why d/acc might be the most coherent philosophy for navigating this transition.
The AI Triad: What I Learned About AI and National Security from My BlueDot Course
Three things make modern AI possible: algorithms, data, and computing power. Understanding how they interact is becoming one of the most important things a policymaker can know.
Continue ReadingOntology in AI: What Does a Machine Think Exists?
Before an AI can reason about the world, it has to decide what the world is made of. That decision is called ontology, and it shapes everything the system can and cannot do.