Hacker News Reader

Comment by professorTuring | Hacker News Reader

The real problem here is when the society will allow a machine to diagnose them and if the society is ready to believe that most diagnostics are probabilistically made.

Up to date we allow humans to be at a 70% error level without problems, but we ask machines to be 100% effective.

The very same happens with autopilot, the big numbers say they drive better than humans but...

View on HN

I remember seeing a statistical analysis here on HN that said the numbers for Tesla autopilots are neither great compared to drivers of Teslas nor do they seem to be fair. (They found a case where a human driver had a crash in what would have been counted as 0 miles in the analysis, indicating that something is inflating the "crashes per miles" metric)

Using autopilot in parking?

This isn't about group think. It is when individuals will allow diagnosis. If you give a woman two options, diagnosis by machine with a record of 80% or a human with a record of 70%, it is a really easy decision to make. The desire to not suffer cancer is strong enough to override almost all emotional arguments. And if you can afford it, you will likely choose both or get a second opinion.

These type of algorithms don't give a confidence interval for their predictions so I don't believe these diagnostics are base on probability at all.

Having a confusion matrix for what the model predict correct or not is not the same as having a CI for the model's prediction.

Isn't it possible to derive a CI from the confusion matrix? https://stats.stackexchange.com/questions/363382/confidence-... or using bootstrap? The authors of this research also provide CIs.

Reply on news.ycombinator.com