
Richard F. answered 03/12/20
Advanced intellectual mentoring from a novelist and philosopher.
Hi Tomy,
Gabe is correct.
However, as a philosopher, I can't resist pointing out what a misleading sentence it is anyway!
Suppose the algorithm says I'm high-risk for doing X but the judge ignores the algorithm and sets me free. OK, so now ...
(a) If I do X, we're tempted to say "Aha! There's the evidence we were looking for! The algorithm 'correctly predicted' my future." Not at all. It predicted that I had a high probability of doing X, and I did X. This shows nothing (repeat, nothing) about the algorithm's accuracy unless we know much more. To see this, suppose it predicted that my 10 best friends would do X, and my 10 closest relatives would not do X, and the results were ... random? Then we'd have good reason to think the algorithm is totally useless, and predicts nothing (even when correct).
(b) If I don't do X ... same problem in reverse. The algorithm might be a superb predictor, but short of perfect. It could easily be wrong in my case but turn out right in all the others. (And even that wouldn't prove it's a reliable predictor - just give some evidence that it is.)
The bottom line: freedom or no, you can tell nothing from one case about what (if anything!) the algorithm is doing.
I know that wasn't your q but thought you might find it interesting!
Richard
Tomy A.
thank you Gabe and Richard. a lot of things i found interesting in your explanation Richard and i appreciate it .the sentence i did sent was from "hello world" by Hanna fry's book in case you guys might be curious03/13/20