Meta will reportedly soon use AI for most product risk assessments instead of human reviewers

According to a report of Meta plans to move the task of assessing the potential damage of its products far from human examiners, rather by pressing the AI ​​to accelerate the process. The internal documents observed by the publication note that Meta aims to have up to 90% of risk assessments on AI, NPR reports, and plans to use AI journals even in fields such as risks for young people and “integrity”, which covers violent content, disinformation and more. Currently named and ancient-named meta-employed that spoke with NPR AI has warned that AI can ignore the serious risks that a human team could have identified.

The updates and new features of Meta platforms, including Instagram and WhatsApp, have long been subjected to human criticism before hitting the public, but Meta would have doubled on the use of AI in the past two months. Now, according to NPR, Product teams must fulfill a questionnaire on their product and subject this for examination by the AI systemwhich generally provides an “instant decision” which includes the risk areas it has identified. They will then have to meet the requirements it has presented to solve the problems before the product can be published.

An old Meta Executive said NPR This reduction in the exam “means that you create higher risks. Negative externalities in product changes are less likely to be avoided before starting to cause problems in the world”. In a declaration at NPRMeta said it would always exploit “human expertise” to assess “new and complex questions” and would leave “low -risk decisions” to AI. Read the full report on .

It happens a few days after Meta took out her – the first since And Earlier this year. The amount of content withdrawn has decreased without surprise following changes, depending on the report. But there was a small increase in intimidation and harassment, as well as violent and graphic content.