By: Jerry Wang
Recent robots using AI created by CLIP, a premade AI model, were used to scan blocks with faces. These robots classified the faces into groups such as “doctor” or “homemaker” in a potentially sexist and racist way.
This shocking result is a result of the way software is made. A supply chain management professor, Zac Stewart Rogers, explained that “[w]ith coding, a lot of times you just build the new software on top of the old software. So, when you get to the point where robots are doing more … and they’re built on top of flawed roots.”
In this case, the algorithm classified black men 9 percent more as criminals compared to white men. It also chose Latino men 6 percent more than white men as janitors and identified women less than males as doctors.
Despite the progressive nature of this issue, so far it has gone nearly unnoticed with most people still seeing robots as impartial. Policy Institute leader, Mauri, Even said “[i]n a legal setting, AI will usher in a new, fairer form of digital justice whereby human emotion, bias and error will become a thing of the past.”
Even in this rapidly growing field, this misconception of robots’ objectivity are able to disseminate due to the way we currently use robots. Robots are currently not used during interpretable tasks. One way we are using robots is to move items in a storage. This would mean that the robot is only using the most efficient path.
However, many companies plan on making robots more suited to abstract tasks. Tesla’s Optimus and Honda’s Asimo are some examples. With the end goal of performing routine and mundane tasks, a prejudice can become an issue. When asked to retrieve an item, these robots may get brands that feature lighter color models or males. Exposing children to stereotypes serves no benefits and could prove to be the most urgent issue with AI.
article: https://s3.amazonaws.com/appforest_uf/f1658069860888x666689010042064400/Robots%20trained%20on%20AI%20exhibited%20racist%20and%20sexist%20behavior%20-%20The%20Washington%20Post.pdf