4 ms·
Along a slightly different track than you were originally running on, there are a number of ways in which an algorithm could promote prejudice. Consider a nice
by kaitai 13y ago
Along a slightly different track than you were originally running on, there are a number of ways in which an algorithm could promote prejudice. Consider a nice little machine-learning algorithm that analyzes characteristics of successful employees at a given company in order to select new hires that share those characteristics. If, for instance, the company had a couple rabid barbecue-lovers who made sure that all lunches primarily featured delicious pork ribs, and they made fun of the vegetarians, Jews, Muslims, and people with braces who couldn't chew the meat off the ribs, and so being a rabid lover of pork was strongly correlated with success in the company and being a non-pork-lover or hampered by braces was strongly correlated with simply leaving the company at an early stage to get a job with the rival across the street that had vegetarian and shrimp satay lunch options... well, it's just the algorithm at work, my friend! Math is truth! As a mathematician, I guarantee it.
Time for dinner. heh.
- vezzy-fnord 13y agoOnce again, this isn't the algorithm being inherently prejudiced. It does what it's meant to do, but it's being applied to a prejudiced and unreliable data set. Don't blame the algorithm for the fallacies of the people using it. The only fault in your scenario is the PEBCAK.