By: Derick Zhang
Unless you live under a rock you would know that AI is becoming a huge part of our lives. But now scientists and engineers are finding out that a robot made to find a criminal telling from their faces on a block, was found out to be racist and sexist.
How scientists and engineers found out was quite self-explanatory; they tested the robot, they tried to find criminals, how the machine worked was they would see the block and use AI to tell if they were a criminal, then put the block into a “criminal” container. They found that they would repeatedly choose a black man’s face. Now they tried “homemaker” and “janitor,” the machine chose blocks with women and people of color. Black people were chosen as “criminal” 9 percent more often than a white man and for “janitor” Latinos were chosen 6 percent more than White men. Even women were less likely to be identified as “doctors” than men. The reason why they are racist, and sexist is because if data put into the robot is biased the output will be biased and they were fitted with billions of images and text captions from the internet and they were biased, and this could have real world implications. For example, if a kid told a robot to get a doll for the kid it would give a white doll. It’s nearly impossible to fix, but that doesn’t mean companies should give up. They should fix the algorithms they use and diagnose the ways they show bad behavior, making way to fix these flaws.
This was alarming and what’s truly horrifying is that billions of dollars are being poured into robotics and it isn’t going to stop. We are currently in a robotics “gold rush” and many wouldn’t even notice how flawed AI really is. Will it be noticed and fixed? If it doesn’t, will we really be overrun by robot overlords? Could this be a waste of money? What’s scary is that there is a chance that this might actually happen.