Notes on trust boundaries, agentic security, and executive oversight — written for people who have to make decisions about systems they did not build.
There are two different classes of AI models behind AI-powered cybersecurity, large language models (LLMs) and small AI models. Both offer opportunities…
Read the field noteAI security refers to the measures and practices designed to protect artificial intelligence systems and their associated data from threats and vulnerabilities.…
Read the field noteWhat happens inside the human brain (at the time of ‘learning’) is something non computational. It falls into those class of computing…
Read the field noteWhat if AI can build better and better versions of itself (self improvement)? Can it be self sustaining?
Read the field noteNurse’s Case – Infant in NICU I am only too familiar with the type of expertise and experience that is required of…
Read the field noteIn his book SuperIntelligence, Bostrom lays out a scenario where a superintelligence may not have the risk-reward understanding the humans do. blithely…
Read the field noteCounterintelligence refers to activities conducted by an organization to protect against or thwart the efforts of foreign or hostile intelligence agencies. The…
Read the field noteA Nurse’s Intuition The wide-ranging possibilities and “black box” nature of AI raise the danger of over-exuberance in some people and fear…
Read the field note