3 ms·
Lending agencies in the USA have also had the hammer brought down on them for paying attention to postal codes in some jurisdictions, but I don't think it's bee
by bunderbunder 8y ago
Lending agencies in the USA have also had the hammer brought down on them for paying attention to postal codes in some jurisdictions, but I don't think it's been banned nationally.
Regardless, the spot where surveillance capitalism worries me is that it could serve to undercut any attempts at banning such practices, regardless of how consumer-friendly a legal jurisdiction tries to be in limiting how the data can be used.
Can't use your current postal code to decide what your interest rate will be? That's fine, we'll just buy a dataset that tells us what your favorite restaurants are, and infer where you live from that instead.
Oh, but we won't do it directly, either, because someone would just blow the whistle on that, too. Instead we'll feed it and several hundred or thousand other data points into a machine learning model. And we'll pick the most black box algorithm we can think of, so we can claim plausible deniability.
Or maybe we'll set up several different online loan companies, each of which offers slightly different rates. We won't explicitly say that only certain people are supposed to use one of them, but we will still engineer the client pools by carefully targeting the advertising based on what websites people tend to visit.
I'm coming to the opinion that we should consider this kind of dragnet collection and reselling of data about people to be abusive prima facie. The potential for misuse is just too great, and the ostensibly legitimate uses that people have been able to come up with are typically really lame and uncompelling.
- ssivark 8y agoTo paraphrase Maciej Ceglowski, this is machine learning as a vehicle for for bias laundering.