Here is a copy of my presentation from a private engineering conference, where I was invited to speak about ML/AI security. It contains the history of police, street lights, and many more relevant past events that should inform our immediate future of machine engineering:

Abstract:

As the old saying goes from the digital prairie, if you see yourself as someone of influence just try ordering someone else’s robot around.

Given the automated platforms everyone has running right now, who is really able to account for what their learning machines are doing, let alone why and how they were authorized? And even if an accounting can be done for one or some robots, who is reliably able solve such questions at scale for increasingly ephemeral operations (assuming robots continue to allow us to terminate them)?

This presentation puts forward protocols and procedures for the idea that a foundation of ethics underpins the standards that will be used to create security solutions that both make intelligent machines more secure from us, as well as make ourselves more secure from them.

Copy of Presentation Slides: security-intell-machines.daviottenheimer.pdf (7 MB)