In the second update from Instagram’s Equity Team, formed in September 2020, the company shared new measures taken to eliminate racial bias and discrepancies on the platform.
Launched internally, Machine Learning Model Cards have been developed with the goal of preventing algorithmic bias. “Model cards work similar to a questionnaire…” and “pose a series of equity-oriented questions and considerations to help reduce the potential for unintended impacts on specific communities, and they allow us to remedy any impact before we launch new technology,” reads the update.
Instagram also plans to add more transparency around post reach and restrictions in an effort to address concerns that BIPOC creators are being suppressed or “shadowbanned” on the app. “We also plan to build direct in-app communication to inform people when bugs and technical issues may be impacting their content.”
One such bug occurred early May, affecting posts by activist movements in Colombia, East Jerusalem and posts related to the National Day of Awareness for Missing and Murdered Indigenous Women.
Wanted to provide a bit more information on what happened here. We are so sorry to everyone who was impacted by this. https://t.co/BFxyJfRbSR pic.twitter.com/3J9YGyRYdS
— Instagram Comms (@InstagramComms) May 7, 2021