It’s one thing to joke about Dogecoin. It’s quite another to joke about public safety. — Last week, Business Insider’s Hannah Towey published an article on Elon Musk’s response to a question on ‘adversarial attacks’ during Tesla’s live-streamed AI Day. Hannah reported that “an attendee asked Musk if he is worried about the car’s lack of 3D sensors allowing adversarial attacks.” …

FACT: Edge cases are limitless and model performance cannot be predicted by traditional accuracy metrics. — We’ve all done it. Deployed a model only to have it fail on some obscure situation and dismissing it as merely an edge case that was unlikely to occur in the first place and claim success as it is quickly trained out of the model. Problem solved? Not so fast. …

This is how state actors will hack our AI systems. — Artificial intelligence systems are susceptible to a new class of cyber attack that cannot be protected using traditional cybersecurity methods. This new class of cyber attack is possible because the inputs to AI models hang outside of the cybersecurity perimeter both during training time and at deployment time. Specifically, during…

The challenge is that each of these pillars is an entirely separate discipline. — Most breakthrough technologies are introduced by first-mover startups that prove out market need before the emerging best-of-breeds are scooped up by larger companies in the inevitable market consolidation that follows. Across many sectors in AI this has already occurred and the innovations are already impacting our daily lives in mostly…

It is time to shift the conversation towards what really matters— acceptable use and protection — “No AI currently exists that can outduel a human strapped into a fighter jet in a high-speed, high-G dogfight.” — DARPA, May 2019 Fast forward to August 2020, which saw an AI built by Heron Systems flawlessly beat top fighter pilots 5 to 0 at DARPA’s AlphaDogFight Trials. Time and…

It’s Time to Sound the Alarm about the Vulnerabilities of Artificial Intelligence — There is a new emerging threat to public safety — artificial intelligence is hackable. Like the innovations made possible by the creation of the Internet, whose infrastructure was never built with security in mind, we are at a similar inflection point with artificial intelligence. Artificial intelligence applications are emerging at…