OpenAI: GPT-4 Can Help with Content Moderation Workload

OpenAI has shared instructions for training to handle content moderation at scale. Some customers are already using the process, which OpenAI says can reduce time for fine-tuning content moderation policies from weeks or months to mere hours. The company proposes its customization technique can also save money by having GPT-4 do the work of tens of thousands of human moderators. Properly trained, GPT-4 could perform moderation tasks more consistently in that it would be free of human bias, OpenAI says. While AI can incorporate biases from training data, technologists view AI bias as more correctable than human predisposition. Continue reading OpenAI: GPT-4 Can Help with Content Moderation Workload

Online Safety Act Paused as Ofcom Reports on Net Neutrality

UK watchdog Ofcom has proposed a loosening of the nation’s net neutrality rules so as to not unduly restrict innovation and development. While it is up to government and Parliament to change the law, recommendations from Ofcom — which was created to monitor compliance with net neutrality laws — are influential. “Since the current rules were put in place in 2016, there have been significant developments in the online world, including a surge in demand for capacity,” as well as the rollout of 5G, and the emergence of large players like Netflix and Amazon Prime. Continue reading Online Safety Act Paused as Ofcom Reports on Net Neutrality

Twitter Roiled by Layoff Talk as Deadline for Musk Deal Looms

Employees at Twitter are reeling following revelations that the workforce may face massive cuts in the year ahead regardless of who owns the company. According to documents obtained by The Washington Post, Twitter’s current management plans to trim the payroll by about $800 million, representing nearly 25 percent of the company’s staff. However, Twitter denies that report. Meanwhile, Elon Musk, who is being sued to force consummation of his $44 billion Twitter purchase, is said to be contemplating elimination of three times as many jobs. Continue reading Twitter Roiled by Layoff Talk as Deadline for Musk Deal Looms

Twitch Aims to Remove Channels That Spread Misinformation

In an effort to thwart misinformation, Amazon-owned live-streaming video service Twitch is cracking down on bad actors. “We do not believe that individuals who use online services to spread false, harmful information, have a place in our community,” the company stated. Twitch worked with researchers and experts to identify three characteristics that all bad actors share: an online presence dedicated to (1) persistently sharing (2) widely disproven and broadly shared (3) harmful misinformation topics, such as conspiracies that promote violence. Twitch specified that it will not take action against “one-off” statements containing misinformation. Continue reading Twitch Aims to Remove Channels That Spread Misinformation

TikTok Adjusts Feed to Curb Repetition, Offers Users Control

TikTok is tweaking its For You feed so users won’t be recommended too much of the same content. The move follows a global spate of regulatory hearings that took social media companies to task for promoting content that intentionally polarizes adults and potentially harms children. In addition to “diversifying recommendations” in order to “interrupt repetitive patterns” around topics that may provide negative reinforcement, like “loneliness or weight loss,” the popular ByteDance platform said it is also introducing a new feature that will allow users to avoid specific topics or creators. Continue reading TikTok Adjusts Feed to Curb Repetition, Offers Users Control

Facebook Removes More Fake Accounts and Hate Speech

In Q1 2019, Facebook removed 2.2 billion fake accounts from its popular social platform. That compares to 583 million fake accounts the company deleted in Q1 2018; in Q4 that year, it removed “just more” than 1 billion. Facebook said that “the vast majority” is removed within minutes of being created, so they do not count in its monthly/daily active user metrics. In its biannual report, Facebook also said its automated detection software used to delete “illicit content” was improving, removing more than half of the targeted speech. Continue reading Facebook Removes More Fake Accounts and Hate Speech