We have many studies that show vulnerabilities of these image classification models to noise (adversarial or otherwise). The confidence levels are just not enou

We have many studies that show vulnerabilities of these image classification models to noise (adversarial or otherwise). The confidence levels are just not enough to fit into our "innocent until proven guilty (beyond reasonable doubt)" justice system. The danger is that big tech/AI can pump enough money into politics that justice system parameters will be rewritten.