Algorithmic filters amplify privilege

Type into almost any AI image generator the phrase “beautiful female holding a flower”, and you are 90% likely to get a Caucasian female.

Algorithmic filters amplify privilege, filter out diverse creators, and suppress dissent.

One of the algorithm’s jobs is to find the most popular answers. If the most popular restaurant in town sells burgers and that’s the one with the most reviews, and you ask. “What’s good for dinner?”, it’s going to give you that burger joint. 

It might ignore the smaller-volume Indian or Thai restaurant that has food that’s just as good.

Algorithmic filters tend to push up what’s already popular, which means anything that diverges from that is much less likely to get seen.

We see this in the popularity algorithmic filters on places like TikTok and Instagram.

Smaller creators struggle to get any recognition simply because the large creators are already sucking up all the oxygen in the room.

Until recently there were a lot of bookstores around, and on the shelves of those bookstores were very often the same 200 authors that it had always been. For example, if you were Stephen King, your book would take up shelf space. That meant that a smaller author who was not yet famous didn’t have a place to put their book. 

Algorithms are doing the same decision making in real time. 

They’re making decisions on who gets that very precious space on your social media feed or on your search engine page based on popularity and other criteria.

This means anything that dissent from that paradigm is much less likely to be seen.

Is there a way for us to change algorithms so that  they include quality, but less popular and less visible, choices?

#AI #AIissues #AIethics #business #AISensum #AIagents #technology #resposibletech #AIreplacementtheory