There’s too many opportunities to leverage AI in ways that we could not imagine, nor want. Recently, I noted an interested experiment in AI driven automated report building by the police. I thought it was interesting, but the ACLU has pointed out some concerns. Their report notes that AI could be driven into directions that include bias, but also into directions which may result in protecting the officer when they stray too far from Constitutional requirements. As I started to think about it, an AI performing in a black box, a non-transparent model, could also look for potential wrong-doing on the parts of anyone being recorded by the police, and then leading towards even more crimes being identified, including by victims who are giving a report that they think is for their own benefit.
A new white paper from the ACLU outlines the risks of generative AI use by America’s police.
Found at mashable.com
