Mark Zuckerberg – CEO Facebook has always been confident on the ability of artificial intelligence (AI) and on more than one occasion claimed that he relies on the technology to remove terrorist content from its platform. However, Facebook’s chief AI scientist, Yann LeCun states that the AI still needs more time to develop, especially when it comes to screening live videos.
Speaking at a recently held event at the social media’s AI Research Lab in Paris, LeCun claimed that even modern platforms are still years away from implementing AI to moderate live videos.
Streaming live videos can be a problematic situation for social media companies if they are mishandled. Facebook also came under strict scrutiny when the recent atrocity at the mosque in Christchurch, New Zealand was streamed live. Although, Facebook claims that fewer than 200 users saw the video, reports reveal that it was downloaded and shared across several internet forums and platforms.
During the event, LeCun pointed out that machine learning system is unable to understand the nuances present in the video automatically but it can remove content that has already been marked ‘inappropriate’ by human.
According to LeCun, the main problem with training AI to moderate ‘violence’ content is the lack of data. Since (thankfully) we don’t see people shooting others in our everyday routine, it is not possible to train systems to recognize such data.
He also clarified that Facebook is using automated systems that are working as assistants to human moderators. When the AI flag illicit content, the humans manually remove it in order to ensure the security of its users.
AI expert, LeCun was recently awarded the Turing Prize along with other AI luminaries.
Also read: Facebook executive deem its platform safe despite having thousands of under 13 users who illegitimately log in to the social network
Photo: AP Photo/Tony Avelar
Read next: Facebook Discloses What Is Cooking In Its Secret Robotic Division
Speaking at a recently held event at the social media’s AI Research Lab in Paris, LeCun claimed that even modern platforms are still years away from implementing AI to moderate live videos.
Streaming live videos can be a problematic situation for social media companies if they are mishandled. Facebook also came under strict scrutiny when the recent atrocity at the mosque in Christchurch, New Zealand was streamed live. Although, Facebook claims that fewer than 200 users saw the video, reports reveal that it was downloaded and shared across several internet forums and platforms.
During the event, LeCun pointed out that machine learning system is unable to understand the nuances present in the video automatically but it can remove content that has already been marked ‘inappropriate’ by human.
According to LeCun, the main problem with training AI to moderate ‘violence’ content is the lack of data. Since (thankfully) we don’t see people shooting others in our everyday routine, it is not possible to train systems to recognize such data.
He also clarified that Facebook is using automated systems that are working as assistants to human moderators. When the AI flag illicit content, the humans manually remove it in order to ensure the security of its users.
AI expert, LeCun was recently awarded the Turing Prize along with other AI luminaries.
Also read: Facebook executive deem its platform safe despite having thousands of under 13 users who illegitimately log in to the social network
Photo: AP Photo/Tony Avelar
Read next: Facebook Discloses What Is Cooking In Its Secret Robotic Division
