> If the threat escalates to break-in level, the vehicle sounds an alarm and begins blasting music
Thanks. Well, on paper this sounds a lot more reasonable than what the parent described, but what they might be experiencing might simply be the difference between marketing speak and the reality of AI systems where the cost of false positives is practically entirely externalized, whereas anger about false negatives gets fed back right to the devs.
Edit: I saw some more reports, this does look like disgusting behavior that Tesla are trying to combat here. Just sad that the solution always seems to be more surveillance.
Thanks. Well, on paper this sounds a lot more reasonable than what the parent described, but what they might be experiencing might simply be the difference between marketing speak and the reality of AI systems where the cost of false positives is practically entirely externalized, whereas anger about false negatives gets fed back right to the devs.