Twitter's ML Ethics, Transparency, and Accountability (META) team has been completely eliminated thanks to Elon Musk's mass firings today. The META team was responsible for examining Twitter's algorithms to ensure they remained fair and unbiased. The team was also responsible for explaining and even publishing Twitter's algorithm–a rare instance of a tech industry giant offering up its code to scrutiny.
As reported by Wired, META director Rumman Chowdhury confirmed she'd been caught up in the mass layoffs, though she seemed almost happy to be out of the company now that Musk is in control. Speaking to the publication, Chowdhury said the latest work at META was to research political bias in order to refine Twitter's penalization systems.
Joan Deitchman, a senior manager on META, confirmed the whole team had been fired, while machine learning research Kristian Lum said it was the entire team "minus one." Either way, the move certainly implies Musk's Twitter won't concern itself with ethical bias or transparency, despite that being one of his stated goals for purchasing Twitter in the first place.
Twitter's META team specifically studied algorithmic bias surrounding race, gender, age, and political viewpoints, with the goal of ensuring a fair and open platform. While ones and zeros are inherently unbiased, the algorithms that data is fed into can be just as flawed as the humans who create them. Many tech companies have invested in ethical AI teams to some degree, with Twitter's reportedly being one of the better ones.
For example, Musk recently stated that Twitter's revenue is down due to leftist activists on the platform, but META recently published a report which stated that right-leaning sources are more often promoted on Twitter than left-leaning ones.
Hilariously, Twitter's own misleading tweet algorithm–something META helped developed–even flagged Musk's tweet as misleading, offering additional context which read: "Reporting shows advertisers suspending or canceling ad buys over concerns with Twitter platform direction, especially as related to content moderation."
Source: Read Full Article