SAN FRANCISCO, CA (March 23, 2020) – The extraordinary measures taken worldwide to limit the spread of the coronavirus disease have disrupted the global economy, as businesses across industries scramble to adapt to a reality few were prepared for. In many cases, companies have stalled operations \u2013 with notable examples including airlines, movie theaters, theme parks, and restaurants among others.<\/p>\n\n\n\n
The disruption facing consumer technology companies like Google, Facebook, Twitter, and others is different. Engagement on social media platforms is unaffected, if not boosted, by the outbreak. However, underneath user trends are significant public relations and revenue risks if content moderation cannot keep up with the volume of user-generated content uploads.<\/p>\n\n\n\n
Hive, a San Francisco-based AI company, has emerged as a leader in helping platforms navigate the disruption through a combination of data labeling services at scale and production-ready automated content moderation models.<\/p>\n\n\n\n
Hive operates the world\u2019s largest distributed workforce of humans labeling data, now more than 2 million contributors from more than 100 countries, and has been able to step in to support emergent content moderation data labeling needs as contract workforces of business process outsourcers (BPOs) have been forced to go on hiatus given their inability to work from home. Further, Hive\u2019s suite of automated content moderation models have consistently and significantly outperformed capable models from top public clouds, and are being used by more than 15 leading platforms to reduce the volume of content required for human review.<\/p>\n\n\n\n
It is no secret that major social platforms employ tens of thousands of human content moderators to police uploaded content. These massive investments are made to maintain a brand safe environment and protect billions of dollars of ad revenue from marketers who are fast to act when things go wrong<\/a>.<\/p>\n\n\n\n Most of this moderation is done by contract workers, often secured through outsourced labor from firms like Cognizant and Accenture. Work from home mandates spurred by COVID-19 have disrupted this model<\/a>, as most of the moderators are not allowed to work from home. Platforms have suggested that they will use automated tools to help fill the gap during the disruption, but they have also acknowledged that this is likely to reduce effectiveness and to result in slower response times than normal.<\/p>\n\n\n\n Hive has emerged in a unique position to meet emergent needs from social media platforms.<\/p>\n\n\n\n As BPOs have been forced to stand down onsite content moderation services, significant demand for data labeling has arisen. Hive has been able to meet these needs on short notice, mobilizing the world\u2019s largest distributed workforce of humans labeling data, now more than 2 million contributors sourced from more than 100 countries. Hive\u2019s workforce is paid to complete data labeling tasks through a consensus-driven workflow that yields high quality ground truth data.<\/p>\n\n\n\n \u201cAs more people worldwide stay close to home during the crisis and face unemployment or furloughs, our global workforce has seen significant daily growth and unprecedented capacity,\u201d says Kevin Guo, Co-Founder and CEO of Hive.<\/p>\n\n\n\n Among data labeling service providers, Hive brings differentiated expertise to content moderation use cases. To date, Hive\u2019s workforce has labeled more than 80 million human annotations for \u201cnot safe for work\u201d (NSFW) content and more than 40 million human annotations for violent content (e.g. guns, knives, blood). Those preexisting job designs and workforce familiarity has enabled negligible job setup for new clients signed already this week.<\/p>\n\n\n\n Platforms are also relying on Hive to reduce the volume of content required for human review through use of Hive\u2019s automated content moderation product suite. Hive\u2019s models – which span visual, audio, and text solutions – have consistently and significantly outperformed comparable models from top public clouds<\/a>, and are currently helping to power content moderation solutions for more than fifteen of the top social platforms.<\/p>\n\n\n\n Guo adds, \u201cWe have ample capacity for labeling and model deployment and are prepared to support the industry in helping to keep digital environments safe for consumers and brands as we all navigate the disruption caused by COVID-19.\u201d<\/p>\n\n\n\nHow Hive is Helping<\/h2>\n\n\n\n