I’m here for this. Finally a positive feedback loop where people are discouraged from illegally parking, transit becomes more efficient, and SEPTA gains a new funding source
I like the idea in theory, and it’s one of the seemingly rare good uses of AI. My concern is that the bill doesn’t have any requirement for outside auditing of the system, and as far as I’ve seen, the data use/storage policies are nowhere to be found for public review.
I emailed my councilperson about this, but haven’t gotten a response yet.
My best guess as an AI engineer is that the AI is to detect the difference between cars that are parked vs cars on the road that are stopping at a light and then take the picture at the right time.
I don’t think the article says the bus drivers initiate the photos — presumably SEPTA would rather have drivers drive than framing the perfect shot — so it makes sense to have AI fit in there.
It’s not “are you recording” but more highlighting what to look at. It would take so many union labor paid hours to sift through all those videos to find a violation. It’s probably cheaper, as crazy as it sounds, to train an AI Model to catch the violations and log it so an employee only has to sift through likely violators.
I also wouldn’t be surprised if the PPA is looking into this model themselves for… other reasons.
Yeah, this is basically just giving an AI pattern recognition service access to camera feeds that are already there. And from what I’ve read, cars that are flagged at violators will be reviewed by a person before the system actually issues a ticket. If the privacy aspects are handled appropriately (big if there), this is probably going to be a pretty good system, I think.
Add comment