https://www.thesun.co.uk/motors/5393978/insurance-race-row-john-mohammed/
Obviously this needs some further details as I'm not sure how much I'd trust the sun to investigate something like this well but ostensible this is rather dubious and awkward for the firms to say the least! If someone's name is affecting a pricing model then they probably have got some rather awkward questions to answer going forwards, especially if regulators take an interest in this. It is a bit surprising if true as I'd previously assumed the insurance industry was quite conservative and rather old school with their models and I'm surprised someone's name is even looked at for this purpose*.
If true then this perhaps is one of the downsides of the whole "big data" thing, you can end up with potentially quite discriminatory results (also a current research topic in machine learning - assessing the "fairness" of algorithms) - for example google classifying black people as gorillas, or the dubious differences in search results when searching for pictures of white people vs pictures of black people etc...
(*partly why I'm still wondering if there is some other explanation for the sun's results)
MOTORISTS are being stung for almost £1,000 more to insure cars if their name is Mohammed.
Top firms Admiral, Marks & Spencer, Bell, Elephant and Diamond all give far lower quotes when the driver has the English name John, a Sun investigation reveals.
The most shocking example we found was an Admiral quote via GoCompare. We put in identical details apart from the name.
When it was “John Smith” wanting fully comprehensive insurance for a 2007 Ford Focus in Leicester the quote was £1,333.
Obviously this needs some further details as I'm not sure how much I'd trust the sun to investigate something like this well but ostensible this is rather dubious and awkward for the firms to say the least! If someone's name is affecting a pricing model then they probably have got some rather awkward questions to answer going forwards, especially if regulators take an interest in this. It is a bit surprising if true as I'd previously assumed the insurance industry was quite conservative and rather old school with their models and I'm surprised someone's name is even looked at for this purpose*.
If true then this perhaps is one of the downsides of the whole "big data" thing, you can end up with potentially quite discriminatory results (also a current research topic in machine learning - assessing the "fairness" of algorithms) - for example google classifying black people as gorillas, or the dubious differences in search results when searching for pictures of white people vs pictures of black people etc...
(*partly why I'm still wondering if there is some other explanation for the sun's results)