Lets say I have a self driving taxi company. I need to prove to the law enforcement that my cars do not break any traffic rules.
The cars ride on urban streets where lots of things going on. I imagine the environment is highly chaotic. The cars are constantly streaming me and the authorities all the data from all their sensors. And lets assume that police do not have any other real world data.
Question is: Can the authorities conclude that I am sending them the correct information, just due to the reasoning that if I were sending them fake data, ie. simulations of an imaginary street that do not contradict physical laws, I need to have an enormous amount of compute?
Police can store large amount of data and they have a lot more time to check whether something is wrong with the data before having to delete them. But if I am sending false information, I have to constantly produce it and send it. This would require a lot more compute than just observing the environment and sending the related information.
Am I right or wrong? Thanks for reading.
