Impromptu: Amplifying Our Humanity Through AI No matter how difficult the conversation, I believe it would be negligent not to discuss the ways in which AI can be used to give us more justice, not less. We will never get it perfect—there will always be some biases inherent in any human system—but we must keep striving toward better. Where things can go wrong The problems with any existing or potential deployment of AI in a criminal justice setting are well-documented. Just the concept has such dystopian potential that it has inspired mul- titudes of sci-fi movies and books—but you don’t need to look to fiction to see where AI usage in criminal justice can and has gone awry. Critics of predictive policing, for example, fear that human bias will be “baked in”—which has definitely already happened. This bias is then reified by the AI algorithm’s seeming authority, which, even more than a judge, might appear to be an “objec- tive” decision maker. Facial recognition algorithms have also been problematic, resulting in multiple studies showing higher false positive rates on darker-skinned individuals. I could go on, but my point here is not to dwell on the down- sides, whether potential or actual. Instead, I think one of the most promising things about GPT-4, and the way OpenAI designed it to be used, is that it is putting AI’s power into the hands of individual citizens in the hope of empowering them in grassroots, decentralized, demo- cratic ways. 68