Sometimes, there's a case for opaque algorithms so that it is harder to game it. For sentencing & parole? No. BASIC RIGHT to cross-examine.
-
-
-
Fundamental violation of right of confrontation in the Sixth Amendment. Yes, I am an ex-Prosecutor...
- Show replies
New conversation -
-
-
@alevin that is a clear violation of the sixth amendmentThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
nope, as fingerprint analysis is challengeable in court and breathalysers need a 2nd sample for confirmation
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Courts/judges can be biased; it may be a good idea to carefully add computation to balance human biases. But not like this, no.
-
This is a good way of framing it. If the code were open source and methods of data collection/scrubbing audited, it could be ...
End of conversation
New conversation -
-
-
Many are. This stuff isn't. Check this. Why did it take Propublica to run super basic tests on the system? https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
IANAL, but wouldn't this have to be a Brady disclosure if used in sentencing? I guess that's why court tossed, but seems obvious?
-
I misread, should be "referred" instead of "tossed."
- Show replies
New conversation -
-
-
and with a AUC (nominally, accurracy) of no better of 0.75, it's wrong at least 25% of the time. http://www.northpointeinc.com/downloads/research/Creating-Risk-Scores-in-Very-Imbalanced-Datasets.pdf … (pdf link)
-
though probably not right to call AUC of an ROC curve accuracy, it does call into question how sensitivity and fall-out are balanced
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.