Surge AI CEO explains why he hates the term 'data labeling'

In a recent podcast, Edwin Chen, the CEO of Surge AI, expressed his strong aversion to the term 'data labeling,' arguing that it oversimplifies the complex work being done in the field. Chen, who has an extensive background in tech giants like Google, Twitter, and Meta, believes that many people misunderstand the intricacies involved in data labeling. 'When people think of data labeling, they often imagine trivial tasks, like tagging cat pictures or outlining cars,' Chen explained during his conversation with Lenny Rachitsky on the 'Lenny Podcast.' Surge AI, founded by Chen in 2020, is a competitor in the AI data labeling market, standing alongside firms such as Scale AI and Mercor. The company has notably partnered with Anthropic and operates DataAnnotation.tech, a platform that allows freelancers to contribute to AI model training. These freelancers, often labeled as 'ghost workers,' play a vital role in the AI development process, working behind the scenes to ensure quality and accuracy. Chen emphasizes that data labeling is not merely about repetitive tasks; it is a fundamentally creative process. He likens the work done by companies like Surge to the nurturing of a child, where teaching values and creativity is as important as imparting knowledge. 'What we're doing is more akin to raising humanity's children,' he remarked, illustrating the depth of responsibility that comes with developing AI. The philosophy behind Surge is reflected on its website, which poses the thought-provoking question: 'What made Hemingway, Kahlo, and von Neumann extraordinary?' The answer highlights the importance of experiences—love, loss, and choices—that shape intelligence, much like how data evolves AI. Reflecting on his journey, Chen shared that he was initially apprehensive about transitioning from a tech role to entrepreneurship, fearing he would be consumed by financial reports and meetings. However, he found that he could continue engaging with data, stating, 'I never had to stop burying my head in the data.' He advised aspiring entrepreneurs not to feel pressured to conform to a business mold, asserting that success can be achieved by focusing on creating exceptional products. 'You don't need to become someone you're not,' Chen concluded, expressing his wish he had known this sooner. For those involved in data labeling or interested in the field, Chen's insights offer a refreshing perspective on the significance of their work in shaping the future of AI.