Hao Zhang

The ideas this documentary conveys aren't bad but as others point out - it could be less emotional. It is very disappointing to see the documentary contains various biased options on China and Machine Learning while it is trying to address the problem of bias. Without involving too much about politics, let's just talk about machine learning. The movie briefly explained how AI/Machine Learning works and made the audience feel the decisions of machines control our lives. This sounds like a very stereotypical impression of AI/ML and I wouldn't expect people with Ph.D. degrees will have such a limited point of view, not to mention that they didn't realize supervised learning is just a subset of matching learning when they try to explain the concepts to audience. To be honest, at the end of the movie, I was more confused than at the beginning. The definition of the kind of "bias" is talking about is unclear to me. Why face recognition is evil and should be entirely banned while vehicle license plate cameras and red light cameras may be allowed? To what extent should our algorithms be "unbiased": should we require ChatGPT to provide the same level of service to people using different languages? These kinds of questions are unanswered after watching it. Finally, the approach they take - simply banning the usage of new technologies they believe contain "bias" surprised me. Even in the movie, it says IBMs quickly improved their model and eliminated the "bias". So clearly it is a problem we could solve. Even in the movie, it implies the problem very likely is in the training data (driving license database). If the majority of the training data is from a specific group of people, the trained model will overfit that group. It is not black magic or anything we don't know how to fix. In summary, the movie talked about how a classic problem exists in new technologies and provided a solution which is to ban the usage of new technologies to avoid having such problems.