The use of an AI-enhanced film that was filed as evidence in a triple murder case in King County, Washington, has been denied by Superior Court Judge Leroy McCullogh. There are two main problems with using these videos in a criminal prosecution. Firstly, AI image enhancement adds details to the original image and subtracts them. Secondly, there is currently no way to demonstrate that the AI adjustments produce a video that faithfully captures the real scenario.
Artificial intelligence (AI) technology has developed swiftly to the point where it may be used to enhance and repair damaged photos. AI is able to accomplish this by producing details through training on millions of previously taken photos. In other words, artificial intelligence (AI) uses the most likely grassy field photos it has previously seen to try and patch in a higher quality replacement when it detects a region in a low quality image that looks like a grassy field. Crucially, since generative AI cannot distinguish between what is genuine and unreal, the improved videos may include or even exclude important details.
Generative AI uses the input of millions of images combined with algorithms to create a massive set of numbers that represent those images. The original images are not kept by the AI, so exactly how AI enhances an image with details it has seen uses “opaque methods to represent what the AI ‘thinks’ should be shown”, according to Judge McCullogh.
