- BRIGHT KENYA TV
- STORY SERIES
- TRENDING MEMES
- FOOD & RECIPE
An internal study by Twitter reveals prejudice in its amplification algorithm in favor of right-wing politics in six countries. The study group consisted of seven countries, and in all of them, aside from Germany, tweets from ‘political right’ accounts received significant exhibition from the algorithm.
The two-part study consisted of Canada, France, Germany, Japan, Spain, and the United Kingdom. Millions of tweets from these countries, posted by elected officials, were evaluated to ascertain whether tweets were amplified in the algorithmically created timeline compared with the reverse-chronological feed.
The second study sought to ascertain whether its amplification algorithms were biased in their recommendations of political content from news channels and feeds. According to the tests, left-wing news feeds and outlets showed less amplification on Twitter feeds than right-leaning outlets. The head of the companyâ€™s machine learning, ethics, transparency, and accountability team, Rumman Chowdhury, claims the bias is present, but that the cause is yet to be accounted for.
The study found that algorithmic recommendations were present from a group standpoint as compared to individual feeds. Thus, individuals in similar political parties wouldn't necessarily receive similar amplification results. The company claims the amplification should not cause concern, claiming all algorithms amplify. However, this scenario brings ethical questions and considerations about how social networks can be utilized to promote certain political content. The company plans to use this data to further research into social media algorithms and socio-political consumption.