3 ms·
Scunthorpe problem [1] is used to describe the false positives for auto filters, which are often results of naive substring matching. In a way, the current prob
by crvdgc 5y ago
Scunthorpe problem [1] is used to describe the false positives for auto filters, which are often results of naive substring matching. In a way, the current problem is similar, but on the semantic level.
However, it doesn't cover other cases for other AI mistakes you mentioned, like self-driving.
[1]: https://en.wikipedia.org/wiki/Scunthorpe_problem https://en.wikipedia.org/wiki/Scunthorpe_problem
- runarberg 5y agoThe problem also goes in the other direction. That is platform rely too much on automation that human signal gets too faint. For example humans have a hard time flagging actual hate speech on these platforms as well. Another example: Every so often there is a front page post on HN where a company like Google will automatically shut down service for a customer (false positive). The customer has a hard time getting through and having this false positive corrected because their signal can’t reach through the layers of automation.