Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I linked my lecture about how I see "ethics" as another term for "politics" in the academic sense.

Other points to note: All AI will always have bias; a) because neural networks literally have a constant called 'bias' in them; and b) training a prediction algorithm means it is being trained to be bias and running a clustering algorithm means lumping commonly themed attributes together.

If you mean "BIAS" as in "RACE/ETHNICITY/SOCIOECONOMIC_STATUS", then most groups already do this, state they do it, and still deal with the general public not believing them.



Reread my last sentence - allow me to rephrase it:

If you mean "BIAS" as in avoiding "RACE/ETHNICITY/SOCIOECONOMIC_STATUS", most groups state they do not use them for decision making and still deal with the "general public" not believing them because results do not support their opinions.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: