OpenAI wants GPT-4 to solve the content moderation dilemma

OpenAI is convinced that its technology can help solve one of the most difficult problems in technology: content moderation at scale. GPT-4 could replace tens of thousands of human moderators while being nearly as accurate and more consistent, OpenAI claims. If that’s true, the most toxic and mentally demanding technological tasks could be outsourced to machines.

in a blog post, OpenAI claims that it has already been using GPT-4 to develop and refine its own content policies, tag content, and make decisions. “I want to see more people operating their trust and security, and moderation (in) this way,” OpenAI’s head of security systems, Lilian Weng. said traffic light. “This is a very good step forward in the way we use AI to solve real-world problems in a way that is beneficial to society.”

OpenAI sees three main benefits compared to traditional content moderation approaches. First, it claims that people interpret policies differently, while machines are consistent in their judgments. Those guidelines can be as long as a book and constantly changing. While humans need a lot of training to learn and adapt, OpenAI argues that large language models could instantly implement new policies.

Second, GPT-4 can supposedly help develop a new policy in a matter of hours. The process of writing, tagging, gathering feedback, and refining often takes weeks to several months. Third, OpenAI mentions the well-being of workers who are continuously exposed to harmful content, such as videos of child abuse or torture.

After nearly two decades of modern social media and even more years of online communities, content moderation remains one of the toughest challenges for online platforms. Meta, Google and TikTok rely on armies of moderators who have to review terrible and often traumatizing content. Most of them are in developing countries with lower wages, work for outsourcing companies, and are fighting mental health as they receive only a minimal amount of mental health care.

However, OpenAI itself is highly dependent on click workers and human labor. Thousands of people, many of them in African countries like Kenya, annotate and tag the content. Texts can be disturbing, work is stressful, and the pay is bad.

While OpenAI touts its approach as new and revolutionary, AI has been used for content moderation for years. Mark Zuckerberg’s vision of a seamless automated system has yet to come to fruition, but Meta uses algorithms to moderate the vast majority of harmful and illegal content. Platforms like YouTube and TikTok have similar systems, so OpenAI’s technology could appeal to smaller companies that don’t have the resources to develop their own technology.

All platforms openly admit that perfect content moderation at scale is impossible. Both humans and machines make mistakes, and while the percentage may be low, there are still millions of harmful posts getting out and plenty of harmless content being hidden or removed.

In particular, the gray area of ​​deceptive, erroneous, and aggressive content that is not necessarily illegal poses a major challenge for automated systems. Even human experts struggle to tag such posts, and machines frequently get it wrong. The same applies to satire or images and videos documenting crime or police brutality.

In the end, OpenAI could help address a problem that its own technology has exacerbated. Generative AI, like ChatGPT or the company’s image creator, DALL-E, makes it much easier to create large-scale misinformation and spread it on social media. Although OpenAI has promised to make ChatGPT more truthful, GPT-4 still willingly produces news-related falsehoods and misinformation.