4 ms·
Surprised not to see mention of the fundamental modern tech problem with regard to insurance. As soon as you have big enough data, and artificially intelligent
by frabcus 9y ago
Surprised not to see mention of the fundamental modern tech problem with regard to insurance.
As soon as you have big enough data, and artificially intelligent enough algorithms, the insurance becomes too predictive.
The whole point of insurance is to pay people when rare, bad events happen to them. If an insurer can predict well enough who will be the victims, it can refuse them insurance, and hence remove the entire purpose and benefit of the insurance.
This is the flip side to moral hazard. What does it even mean to offer commercial insurance, if it is only offered to the people who least need it?
This is least bad, for example, with a car predicting you're a bad driver as you can improve behaviour. It is really problematic with data such as gene sequencing, which you can't do anything about.
Only way out I can see is compulsory insurance, levied as a tax. Or maybe non-profit or Government AIs, trained to find a sweet spot between moral hazard and its opposite?
- xyzzyz 9y agoOnly way out I can see is compulsory insurance, levied as a tax. I cannot help but notice that you basically reinvented welfare state.
- Kluny 9y agoOnly if it's applied as a blanket policy that insures everything and everyone has to pay it. In a more limited way, you can look at British Columbia's Autoplan insurance. It's a crown corporation that provides insurance to everyone in BC. Everyone is required by law to carry insurance on their vehicle, and they have to buy it from ICBC. It's not the cheapest in the world, as they have no competitors, but they are legally required to keep costs fairly low. So everyone in BC can get car insurance and it ends up costing something like $50-150 per month per car. No one gets denied, no one goes without coverage, and they have a big enough customer base to cover their costs.