Meta is still active on changes that were released last year

One year later In its first research on human rights, Meta says it is still working on a number of changes that analysts endorse. The company was liquidated to describe in detail how they are doing in dealing with the many opinions of readers.

According to the company, it has already set 65 points out of 117, while another 42 have been labeled “ongoing or ongoing.” However, there are six areas in which the company says it still wants “possible” reforms and two ideas that the company has “refused” to take action. And, in particular, some of these are tackling the challenges posed in the 2020 initial assessment.

The original report, released in July 2020, found that the company needed to do more to stop “pushing users into magical rooms.” It also said the company should address issues related to algorithmic bias, and criticized the company’s handling of Donald Trump’s policies. In his , Meta states that it has not yet implemented all the changes that researchers have sought regarding algorithmic bias. The company has made some changes, such as partnering with external experts and adding diversity to its AI team, but says some changes are “under control.”

In particular, the researchers called for a comprehensive, company-wide approach to “prevent, identify, and address potential causes of discrimination and discrimination in the production or deployment of AI and machine learning systems” and to “constantly test existing algorithms and machine-learning models”. “Meta said the idea” has been reviewed. ” Similarly, the study also recommended “lessons learned in understanding and minimizing the sources of prejudice and discrimination in AI for all groups that develop algorithms and machine learning systems.” Those ideas were also listed as “illuminated,” according to Meta.

The company also said that some of the regulatory changes that are being “are being monitored.” This includes ideas to promote the “transparency and consistency” of decisions about reduced grievances, and the process for the company to learn more about how hate speech spreads, and how to use data to address hate that needs to be addressed quickly. The researchers also recommended that Meta “reveal more” what users are being monitored for and voting for by voters on its platform. Those ideas are also “being explored.”

The only two ideas that Meta rejected were actually related to census decisions and procedures. “Investigators have suggested that all reports of user-related incidents of vote-rigging be submitted to the auditors to determine if they are in violation of our rules, and that the grievance mechanism be added to the voter harassment,” Meta wrote. But the company said it chose not to change this because it was slowing down the review, and because “many of the alleged voting irregularities do not violate company policy.”

Separately, Meta also claimed to be “The process of learning our platforms and realizing the opportunity to increase justice when it comes to competition in the United States.” To achieve this, the company conducts “platform surveys” and analyzes the use of names and zip code.