According to Engadget, Twitter announced a new plan to study the fairness of its algorithm. As part of this effort, which the company calls the “Responsible Device Learning Program”, data scientists and engineers from across the company will study the potential “unintentional harm” caused by its algorithms and make the results public.
“We are conducting in-depth analysis and research to assess whether the algorithms we use are potentially harmful,” the company wrote in a blog post announcing the initiative. First, the company will study Twitter’s image cropping algorithm, which has been criticized for favoring people with whiter skin. Twitter will also study its content recommendations, including “evaluation of the fairness of our Home timeline recommendations in different ethnic subgroups” and “analysis of content recommendations of different political ideologies in seven countries.”
It is not yet clear how much impact this move will have. Twitter pointed out that in some cases, it may change various aspects of its platform based on its research results, while other research may simply lead to “important discussions around the way we build and apply ML (device learning).
But for Twitter and other social media platforms, this issue is a timely one. After the riots in the U.S. Capitol, lawmakers have put pressure on Twitter, YouTube, and Facebook to make their algorithms more transparent. Some legislators have proposed legislation requiring these companies to assess whether their algorithms are biased.
Twitter CEO Jack Dorsey (Jack Dorsey) also said that he hopes to establish an algorithm “market” so that users can control which algorithms they use. The company stated in its latest blog post that it is in the “early stages of exploration” for similar ideas.