Twitter said Wednesday it would investigate whether its machine learning technology is causing unintended harm. This is a move that is causing social media companies to be increasingly scrutinized for their role in spreading dangerous conspiracy theories and facilitating harassment.
In this photo image, a Twitter logo is displayed on a smartphone and a computer screen … [+]
SOPA Images / LightRocket via Getty Images
In the coming months, Twitter’s Responsible Machine Learning Initiative will publish reports analyzing possible racial and gender biases in their image cropping algorithm, “a fairness assessment of our home timeline recommendations for different racial subgroups” and “a Analysis of content recommendations for different political ideologies seven countries. “
After Twitter’s image algorithm was criticized last year for focusing on white faces versus darker ones, the company claimed that its tests showed no racial or gender bias, although features were subsequently announced that would give more users more control over Enable images because “We Realize that the way we automatically crop photos is potentially harmful. “
According to Twitter, the results of the reports can spark changes to the platform, new guidelines for the design of certain products, and “increased awareness” of ethical machine learning.
“When Twitter uses ML it can affect hundreds of millions of tweets a day and sometimes the way a system is designed can behave differently than intended,” the company said. “These subtle changes can then affect the people who use Twitter, and we want to make sure we research those changes and use them to create a better product.”
The Twitter initiative comes from the fact that social media companies are exposed to allegations that their algorithms are responsible for the increasing polarization, echo chambers, misinformation and online radicalization. Twitter, in particular, has been criticized for not doing enough to combat harassment. Former New York Times columnist Charlie Warzel argued in his Galaxy Brain newsletter this week that the Trending section of Twitter is drawing the attention of the entire Internet to individual users, leading to disproportionate outrage and widespread occurrence “Abort culture” leads. CEO Jack Dorsey previously said he wanted to see a future where users could choose which algorithm to use in an App Store-like interface instead of relying on an algorithm from Twitter.
Twitter’s tacit recognition that its algorithms could be harmful to users and society at large contrasts with Facebook’s defense of its recommendation algorithms. In a Medium post last month, Vice President of Global Affairs Nick Clegg argued that Facebook has no commercial interest in amplifying extreme content and that the platform is not solely responsible for increasing political polarization, as individual choices have an impact on what users see on their news feeds. “We have to look at ourselves in the mirror and not wrap ourselves in the false comfort that we have only manipulated by machines all along,” he said.