These devices are all driven by machine learning voice recognition. I have no idea how I am supposed to derive trust in the software running on them when even t

These devices are all driven by machine learning voice recognition. I have no idea how I am supposed to derive trust in the software running on them when even the engineers themselves are arms-distance away from their machine models, and the machine models are tuned to terabytes of input data and are trained with reinforcement learning. I don't want to disparage engineers working on these products, but hell, we don't even know how big software systems really work anymore, and we've got neural networks thrown in the mix. We're all just plugging shit together until it limps these days. I vote no on telescreens.