A Skeptic's Guide to Hacking AI (Jim Harris)

After a long hiatus, NCAS is pleased to announce our first in-person talk in quite a while!

Virtually every product and service advertised today claims some use of “AI,” but what is AI, how does it work, and how can it be “hacked?” Jim Harris, a Distinguished Engineer and accredited Model Risk Officer, provides a Skeptic’s take on how Large Language Models (LLMs) work, why they fail, and how you can trick them into exposing their weaknesses.

Jim Harris is a board member of NCAS. He graduated in 1994 from North Carolina State University with a BS in Physics. Since that time, he has worked as a hardware engineer for IBM and Silicon Image, a Special Agent for the FBI, a cybersecurity consultant, a tech startup executive, and a Distinguished Engineer and Model Risk Officer for Capital One Financial.

(Note: the location of the talk is the Connie Morella Library in Bethesda. Facebook is still calling it the Bethesda Regional Library, which is its old name)