[HN Gopher] Most users cannot identify AI bias, even in training...
___________________________________________________________________
Most users cannot identify AI bias, even in training data
Author : giuliomagnifico
Score : 12 points
Date : 2025-10-18 18:13 UTC (4 hours ago)
(HTM) web link (www.psu.edu)
(TXT) w3m dump (www.psu.edu)
| DocTomoe wrote:
| If bias can only be seen by a minority of people ... is it really
| 'AI bias', or just societal bias?
|
| > "In one of the experiment scenarios -- which featured racially
| biased AI performance -- the system failed to accurately classify
| the facial expression of the images from minority groups,"
|
| Could it be that _real_ people have trouble reading the facial
| expression of the image of minority groups?
| ecocentrik wrote:
| By "real people" do you mean people who are not members of
| those minority groups? Or are people who can "accurately
| classify the facial expression of images from minority groups"
| not "real people"?
|
| I hope you can see the problem with your very lazy argument.
| eth0up wrote:
| Not to be glib, but...
|
| Grok
| delichon wrote:
| There are two goals in tension: 1) accurate, 2) unbiased. In
| order to be more unbiased we must limit the training data to the
| most available for the least represented subclass we want to
| support. In order to be more accurate we train on all of the data
| we can get, bias intact.
|
| So the choice is a function of both. How much bias or inaccuracy
| can you tolerate? Each use case has a different answer.
| 7e wrote:
| According to research, white Americans report as happier than
| other groups. So I'm not sure there's bias here, only unhappiness
| about that result.
___________________________________________________________________
(page generated 2025-10-18 23:00 UTC)