Listen to Your (Tipsy) Inner Voice: AI Is Not Your Drinking Buddy

Tori R.A. Kricken

AI is everywhere, helping and protecting society, right? In a mere twelve seconds, AI can use a person’s literal voice, regardless of the words spoken, to detect alcohol intoxication and save lives risked by intoxicated behaviors. Recent advances in artificial intelligence have reached a point where this technology not only exists, but also is inexpensive, instantaneous, easily accessible, increasingly accurate, and potentially undetectable. However, this voice analysis can also incriminate subjects and potentially reveal deeply personal information, all while bypassing typical constitutional protections. Such technology should be questioned lest society substitute for human judgment encoded procedures that obliterate privacy and misjudge character—often discriminately so for traditionally underrepresented populations. The necessary companion to “big data” breakthroughs of voice analysis is a thoughtful assessment of how to adapt individual rights in light of these advancements. From First Amendment freedoms through Fourteenth Amendment due process and equal protection, audio-based deep-learning algorithms used to identify alcohol inebriation present constitutional challenges that must be addressed. This Article does just that.