The Untold Worker Exploitation Behind Large Language Models

Scale AI, a company founded by Alexandr Wang, has raised over $1 billion in venture capital, but the story of how the company came to be is less well-known. The company is a massive data-labeling firm that plays a crucial role in the development of AI technology. AI companies need large amounts of data to train their models, and they often turn to the internet as a source of this data. However, this data can contain harmful content, and companies use human workers to label and classify this data to filter out harmful responses. This process is known as Reinforcement Learning from Human Feedback (RLHF). Hundreds of thousands of underpaid workers, mostly from online "cloudwork" platforms, contribute to this process, but they are often mistreated and underpaid. A report by the Oxford Internet Institute highlights the business malpractices of these platforms, including the lack of payment mechanisms, overwork, and poor contracts. Workers on these platforms have reported being traumatized by the content they have to review, and they are not protected by wage laws or provided with counseling. The mistreatment of these workers is a major issue in the development of AI technology, and it needs to be addressed through policy changes and awareness-raising. By spreading awareness of these issues, we can work towards a more responsible and safe development of AI technology.