Twitter Says Its Algorithm Amplifies Right-Wing Content And They Don't Know Why

Twitter claims that its algorithm amplifies right-wing content more than it does left-leaning content, but the social media company doesn't know what's causing this.

Revealed by Twitter in a blog post, the findings were drawn from an internal study of what kind of political content gains traction on Twitter. For some reason, machine learning is leaning towards the right in terms of what Twitter shows as recommendations.

Cleartwo.co.uk

Twitter's right-wing bias

Twitter analysed millions of Tweets posted between April 1 and August 15 in 2020. These Tweets were posted by news outlets as well as by elected officials in countries like France, Canada, Germany, Spain, Japan, the UK, and the US.

In every country except Germany, Twitter found that its algorithms were prioritising right-leaning accounts over the "political left". Owing to this bias, media outlets that publish right-wing content also received more traction.

Even then, Twitter doesn't know why its artificial intelligence is amplifying right-wing content on the platform in so many countries. In the study, Twitter acknowledges the difficulty of answering this question, saying that it is a "product of the interactions between people and the platform."

Twitter CEO Jack Dorsey / Reuters

Also read: Twitter Lets You 'Soft Block' Insufferable Followers: But Will It Limit Abuse?

While there may not be a concrete answer for this phenomenon yet, it may have something to do with Germany's agreement with Big Tech (Facebook, Twitter and Google) wherein companies are required to remove hate speech within 24 hours. In fact, many Twitter users claim to have changed their location to Germany to escape any Nazi imagery from appearing on their screens.

Also read: Twitter Allows Bitcoin Tipping, Becomes 1st Social Platform To Embrace Crypto

Reuters

Regardless, such acknowledgements based on internal studies by social media giants themselves could help fix most of the problem.

Is fast-tracking bias and hate speech like Germany the right way to quell bias on machine learning platforms like Twitter? Share what you think should be done in the comments below.

For the latest in all-things-Twitter and the world of technology and science, continue reading Indiatimes.com.

Citation

Examining algorithmic amplification of political content on Twitter. (2021). Twitter.