3 ms·
“We're not trying to build AI that uses physical features to deny claims”
- version_five 5y agoThis seems legitimate to me - I didn't see the wording of the original tweet but their explanation seems a lot more logical than claiming they are relying on phrenology to analyze trustworthiness of a claimant. That said, I can't imagine ever using a product that asks me to send a video of myself to get service. Maybe I'm just old but it feels dehumanizing, same as job applications that ask for a video. Also, from their clarification on their website: > AI is non-deterministic... That’s why we never let AI perform deterministic actions I don't understand what this means. Do they mean "conclusive"?
- steepben 5y agoAccording to page 128 of their S-1 filing [0], their claims bot "AI Jim" processes claims without human intervention in a third of cases, including automatically denying claims. So either the second tweet in the thread is a lie, or their S-1 filing is a lie. This was pointed out by Rachel Metz [1]. [0]: https://sec.report/Document/0001047469-20-003416/#cy40510_business https://sec.report/Document/0001047469-20-003416/#cy40510_bu... [1]: https://twitter.com/rachelmetz/status/1397625251257753606?s=20 https://twitter.com/rachelmetz/status/1397625251257753606?s=...
- perl4ever 5y agoI think the real story will be that their nature is a financial engineering free lunch that is going to blow up spectacularly. They say they are different because they reinsure most of their risk. So why doesn't everybody do this, print money and let someone else take the risk? Just skimming the S-1 makes me think of Greensill. It's some sort of arbitrage that makes it hard to put your finger on where the risk is.