In this episode of No Priors, Sarah and Elad interview Edwin Chen, the CEO of Surge, a bootstrapped human data startup that has achieved over a billion in revenue. Edwin discusses Surge's focus on high-quality human data for AI advancement, contrasting it with the "body shop" approach of competitors. He shares his experiences working on ML at Google, Facebook, and Twitter, which led him to found Surge to solve the problem of obtaining quality data for training models. The conversation covers Surge's bootstrapping journey, the importance of prioritizing product and problem-solving over fundraising, and the company's approach to measuring and ensuring data quality through technology and machine learning. Edwin also touches on the role of humans in data generation as models become more advanced, the limitations of synthetic data, and the need for external validation to align models with desired objectives. He shares his thoughts on the competitive landscape of frontier models, highlighting the unique strengths of different models and expressing his belief in a diversity of models. Finally, Edwin discusses Surge's plans for public research and education to steer the industry towards better practices and metrics for evaluating model capabilities.
Sign in to continue reading, translating and more.
Continue