Graduating from Data Labeling to AI Data Development

At the core of Snorkel’s approach and platform is the concept of programmatic data development, which we have repeatedly proven to accelerate training of predictive models by leveraging SME knowledge to produce labels efficiently. At Snorkel Research we have been extending this approach to Generative AI, where the output is not just labels but long form and complex. Our strategy has been to develop methods and tools to keep SMEs in the loop of training and evaluation with scalable processes.

In this presentation, Chris Glaze, Principal Research Scientist at Snorkel AI will give a brief history of how Snorkel evolved from data labeling to data development for Generative AI, and share stories on how Snorkel’s applied this approach to Fortune 500 use cases. Angela Fox, Staff Product Designer at Snorkel AI, will demonstrate how users can leverage Snorkel Flow’s unique programmatic labeling features to solve both predictive and generative business challenges.