The social network Twitter said Wednesday that it will launch an initiative on “responsible machine learning” that will include reviews of algorithmic fairness on its platform.
The California-based company said the plan aims to offer more transparency about its artificial intelligence and address “the potential damaging effects of algorithmic decisions.”
The initiative comes amid growing concern over the algorithms used by online services, which some say can promote violence, extremist content and reinforce racial or gender bias.
“Responsible use of technology includes studying the effects it can have in the long term,” Jutta Williams and Rumman Chowdhury, from Twitter’s ethics and transparency team, write in a blog.
When Twitter uses machine learning or machine learning, “it can affect hundreds of millions of tweets a day and sometimes the way a system was designed to help could start to behave differently than expected,” they explained. .
According to the researchers, the initiative seeks to “take responsibility for our algorithmic decisions” with the aim of achieving “fairness and justice in the results.”
“We are also building machine learning solutions that are explainable, so that our algorithms can be better understood, what informs them and how they impact what (the user) sees on Twitter.”
Williams and Chowdhury said the team will share its findings with outside researchers. “To improve the collective understanding of the industry on this issue, help us improve our approach and hold ourselves accountable.”
This Twitter strategy follows a series of controversies in Google’s artificial intelligence ethics team, which have resulted in the firing of two high-level researchers and the resignation of a high-ranking scientist.
